Dual-stack network adaptive scheduling method and system based on DNS resolution strategy

Abstract

This application relates to the field of network communication technologies, and provides a dual-stack network adaptive scheduling method and system based on DNS resolution strategies, for establishing an effective protocol stack feature association mechanism, so as to achieve intelligent scheduling across protocol layers. Among them, the method includes: receiving a DNS resolution request sent by a client, where the DNS resolution request carries a target domain name and a request protocol type; parsing the request protocol type to determine the network protocol stack type supported by the client; generating an adaptive scheduling strategy corresponding to the target domain name according to the network protocol stack type and a preset address type priority list; selecting a target IP address type from a pre-configured IP address pool based on the adaptive scheduling strategy; and returning the resolution result corresponding to the target IP address type to the client to guide the client to establish a network connection through the target IP address type.

Description

Technical Field

[0001] This application belongs to the field of network communication technology, and particularly relates to a dual-stack network adaptive scheduling method and system based on DNS resolution strategy. Background Art

[0002] In the current Internet environment, the network protocol stacks supported by client devices show a diversified development trend. The coexistence of IPv4 and IPv6 dual-stack deployment, as well as the coexistence of QUIC protocol and TCP protocol, is becoming increasingly common. Traditional DNS resolution systems usually rely on statically configured address return policies. For example, they preferentially return IPv6 addresses to promote network upgrade, or allocate IP addresses of different protocol types through a simple round-robin mechanism.

[0003] However, the existing technologies fail to effectively identify the network protocol capabilities actually supported by the client, resulting in the possibility that the returned address type may exceed the compatibility range of the client protocol stack, leading to network connection failures. Secondly, the traditional address selection strategy lacks a dynamic adaptation mechanism and cannot dynamically adjust the priority according to the characteristics of the client protocol stack, resulting in low network resource utilization efficiency. In addition, existing solutions mostly rely on the client to actively report protocol support information, increasing additional communication overhead and implementation complexity, especially prone to latency sensitivity problems in the mobile scenario. It can be seen that the existing DNS systems have not established an effective protocol stack feature association mechanism and are difficult to achieve intelligent scheduling across protocol layers. Summary of the Invention

[0004] This application provides a dual-stack network adaptive scheduling method and system based on DNS resolution strategy to establish an effective protocol stack feature association mechanism, thereby achieving intelligent scheduling across protocol layers.

[0005] In a first aspect, an embodiment of this application provides a dual-stack network adaptive scheduling method based on DNS resolution strategy, which is applied to a dual-stack network adaptive scheduling system. The method includes: receiving a DNS resolution request sent by a client, where the DNS resolution request carries a target domain name and a request protocol type; parsing the request protocol type to determine the network protocol stack type supported by the client; generating an adaptive scheduling strategy corresponding to the target domain name according to the network protocol stack type and a preset address type priority list; selecting a target IP address type from a pre-configured IP address pool based on the adaptive scheduling strategy; and returning the parsing result corresponding to the target IP address type to the client to guide the client to establish a network connection through the target IP address type.

[0006] In a second aspect, an embodiment of the present application provides a dual-stack network adaptive scheduling system, which includes a processor and a memory. Among them, the memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to execute the steps of the above method.

[0007] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, which includes a computer program. When the computer program runs on a dual-stack network adaptive scheduling system, the computer program is used to cause the dual-stack network adaptive scheduling system to execute the steps of the above method.

[0008] The embodiment of the present application realizes multi-dimensional optimization of network connections through a protocol awareness and intelligent scheduling mechanism, significantly improving the resource adaptation ability in a heterogeneous network environment. Different from the passive response mode of traditional DNS resolution, this solution innovatively constructs a closed-loop logic for client protocol stack parsing and address type decision-making, which can deeply identify the network protocol support characteristics of the client and dynamically generate adaptation strategies. By introducing an elastic mapping mechanism for the address type priority list, the optimal IP address type can be intelligently selected according to the real-time protocol stack characteristics, and the client is actively guided to establish the most efficient network channel on the basis of ensuring protocol compatibility. This technology breaks through the limitations of traditional single-dimensional scheduling, realizes seamless coordination between the IPv4 / IPv6 dual-stack environment and emerging network protocols (such as HTTP / 3), and effectively avoids latency losses caused by protocol fallback. At the same time, the pre-decision mechanism based on protocol stack characteristics can significantly shorten the connection establishment time window and improve the end-to-end communication efficiency by eliminating redundant protocol negotiation processes. This protocol-sensitive intelligent scheduling system not only strengthens the forward adaptation ability of network services, but also provides a smooth transition path for the evolution of the multi-protocol ecosystem, with significant network resource optimization value.

[0009] In summary, the embodiment of the present application can establish an effective protocol stack feature association mechanism, thereby realizing intelligent scheduling across protocol layers. Description of the Drawings

[0010] Figure 1 It is a schematic flowchart of a dual-stack network adaptive scheduling method based on DNS resolution strategy provided by an embodiment of the present application.

[0011] Figure 2 It is a schematic structural diagram of a dual-stack network adaptive scheduling system provided by an embodiment of the present application. Detailed Embodiments

[0012] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions of this application in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are some, but not all, of the embodiments of the technical solutions of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments recorded in this application document without creative efforts belong to the scope protected by the technical solutions of this application.

[0013] Refer to Figure 1 , which is a dual-stack network adaptive scheduling method based on DNS resolution strategy provided in the embodiments of this application. This method can be applied to a dual-stack network adaptive scheduling system, and the specific process is as follows in steps 101 - 105.

[0014] Step 101: Receive a DNS resolution request sent by a client, where the DNS resolution request carries a target domain name and a request protocol type.

[0015] In this embodiment, the dual-stack network adaptive scheduling system receives the DNS resolution request sent by the client through the core switch of the enterprise office network. The client is an office terminal device used by enterprise internal employees, which accesses the enterprise local area network through wired or wireless means and triggers the domain name resolution process when accessing the target business system.

[0016] For example, when an employee enters the target domain name "oa.example.com" in the browser, the client generates a DNS query message containing the target domain name and the request protocol type. The request protocol type is implicitly carried through the extended field EDNS Client Subnet (ECS) in the DNS message header, which is specifically manifested as the protocol stack type activated in the client network interface card (NIC) configuration. In the enterprise network environment, after the request passes through the security policy filtering of the edge firewall, it is forwarded by the enterprise internal network DNS server to the resolution processing module of the dual-stack network adaptive scheduling system.

[0017] Step 102: Analyze the request protocol type to determine the network protocol stack type supported by the client.

[0018] In this embodiment, the protocol analysis module of the dual-stack network adaptive scheduling system deeply analyzes the received DNS resolution request, and focuses on extracting the features related to the request protocol type. The analysis process is achieved by deconstructing the application layer payload of the DNS message, which specifically includes identifying the transport layer protocol identifier, parsing the DNS extension option field, and associating network session context information.

[0019] For example, when the client network protocol stack is only configured with an IPv4 address, the generated DNS query message will only contain the IPv4 source address field, and no IPv6-related parameters will be carried in the EDNS extension. If the client has enabled the dual-stack protocol, the DNS request will contain both the IPv6 flow label field and the IPv4 Type of Service (ToS) field. By comparing with the preset terminal protocol configuration list in the enterprise network device management system, the system can accurately determine the actual network protocol stack type supported by the current client, including three states: IPv4 single-stack, IPv6 single-stack, or dual-stack.

[0020] Step 103: Generate an adaptive scheduling policy corresponding to the target domain name according to the network protocol stack type and the preset address type priority list.

[0021] In this embodiment, the policy generation module of the dual-stack network adaptive scheduling system generates an adaptive scheduling policy for the target domain name based on the identified network protocol stack type and in combination with the address type priority list preset by the enterprise network administrator. The address type priority list is stored in the policy database of the system, and its content reflects the protocol evolution strategy of the enterprise network. For example, rules are set such as dual-stack clients give priority to using IPv6 addresses, and single-stack clients match according to the existing protocol type.

[0022] For another example, when the business system corresponding to the target domain name "oa.example.com" has completed the IPv6 transformation, the priority list will be configured as follows: dual-stack clients return AAAA records, IPv4 single-stack clients return A records, and IPv6 single-stack clients trigger the protocol conversion service. During the policy generation process, the dual-stack network adaptive scheduling system also refers to the network status metrics collected in real time, including the link load rate of each protocol stack, the measured end-to-end delay, and the capacity of the address resource pool, and dynamically adjusts the weight parameters of the finally output policy.

[0023] Step 104: Select the target IP address type from the pre-configured IP address pool based on the adaptive scheduling policy.

[0024] In this embodiment, the address selection module of the dual-stack network adaptive scheduling system selects a candidate address set that meets the target address type from the pre-configured IP address pool according to the generated adaptive scheduling policy. The IP address pool is dynamically synchronized by docking with the enterprise network infrastructure management platform and contains the published IPv4 A records and IPv6 AAAA records of each business system.

[0025] For example, for the enterprise file server corresponding to the target domain name "file.example.com", two records, i.e., the IPv4 address 192.0.2.45 and the IPv6 address 2001:db8::1:10, are maintained in the address pool. When the policy requires preferentially selecting the IPv6 address, the selection module will perform the following operations: First, verify the reachability of the IPv6 address in the current network topology. Second, check the service status of the IPv6 port of the target server. Finally, select the optimal access point in combination with the client's geographical location (obtained through EDNS Client Subnet). For scenarios with multiple qualified addresses, the system uses the weighted round-robin algorithm for load balancing, and the weight values are dynamically calculated based on the historical connection success rate and response time.

[0026] Step 105: Return the parsing result corresponding to the target IP address type to the client to guide the client to establish a network connection through the target IP address type.

[0027] In this embodiment, the response processing module of the dual-stack network adaptive scheduling system encapsulates the parsing result corresponding to the selected target IP address type into a standard DNS response message and returns it to the client through the enterprise intranet DNS server. The response message strictly follows the RFC standard format and adapts different transport layer encapsulation methods according to the protocol stack type supported by the client.

[0028] For example, for a client supporting IPv6, the system carries a resource record (RR) of the AAAA record type in the Answer Section of the response message, and its RDATA field is filled with the verified IPv6 address 2001:db8::1:10; for a client supporting only IPv4, a resource record of the A record type is returned. This parsing result guides the client to establish a network connection conforming to the target address type. When the client application layer initiates a TCP three-way handshake, its protocol stack automatically selects the IP version matching the parsing result for data packet encapsulation. The system synchronously updates the parsing log and triggers the connection tracking mechanism to record the address type actually used by the client and the connection establishment success rate, providing data support for subsequent policy optimization.

[0029] In the above application scenario, the dual-stack network adaptive scheduling system realizes the intelligent perception of the protocol stack type and the dynamic scheduling of address resources by deeply integrating the infrastructure of the enterprise network. Thus, it effectively solves problems existing in traditional dual-stack networks, such as rigid protocol selection and low address resource utilization rate. While maintaining backward compatibility, it gradually promotes the IPv6 deployment process of the enterprise network. In addition, through the adaptive policy generation mechanism, the dual-stack network adaptive scheduling system can adjust according to network state changes and business requirements, automatically optimize the traffic distribution of different protocol stacks, and significantly improve the reliability of enterprise network services and resource utilization efficiency.

[0030] In one implementation, the generating, according to the network protocol stack type and a preset address type priority list, an adaptive scheduling policy corresponding to the target domain name in step 103 includes:

[0031] Step 1031: Obtain in real time a first set of real-time parameters of the IPv4 network link associated with the target domain name, where the first set of real-time parameters includes an IPv4 link delay, an available bandwidth of an IPv4 node, and an IPv4 request response success rate.

[0032] In the embodiment of the present application, the link monitoring module of the dual-stack network adaptive scheduling system obtains in real time a first set of real-time parameters of the IPv4 network link associated with the target domain name through the traffic mirroring port of the core switch of the enterprise office network. Among them, the IPv4 link delay is obtained by sending an ICMP probe packet to the authoritative DNS server of the target domain name and calculating the round-trip time. Specifically, in implementation, the dual-stack network adaptive scheduling system targets the monitoring node 192.0.2.45 preset in the IPv4 address pool corresponding to "oa.example.com" and sends a probe message containing a specific timestamp every 5 seconds. The measurement of the available bandwidth of the IPv4 node is based on the traffic statistics information of the enterprise wide area network egress router and is calculated by analyzing the relationship between the change in the TCP session window size and the data packet transmission rate within a unit time. The IPv4 request response success rate is obtained by statistically calculating the TCP three-way handshake success ratio of the client's HTTP connections to the IPv4 address of the domain name in the most recent 10 minutes. For example, if 98 out of 100 connection requests are successfully established sessions, the success rate is recorded as 98%.

[0033] Step 1032: Synchronously collect a second set of real-time parameters of the IPv6 network link associated with the target domain name, where the second set of real-time parameters includes an IPv6 link throughput, an IPv6 node load balancing metric, and an IPv6 path hop count.

[0034] In step 1032, the IPv6 performance collection unit of the dual-stack network adaptive scheduling system interacts with the enterprise software-defined network controller to synchronously collect the second set of real-time parameters of the IPv6 network link associated with the target domain name. The measurement of the IPv6 link throughput adopts the segment routing monitoring technology. For the service cluster where the IPv6 address 2001:db8::1:10 corresponding to "oa.example.com" is located, by analyzing the flow label field in the IPv6 extension header that crosses the enterprise data center border router, the total amount of payloads transmitted within a unit time is statistically calculated. The IPv6 node load balancing metric is obtained by parsing the real-time resource utilization data of the target server, specifically including the weighted average of three parameters: the CPU usage rate, the memory occupancy rate, and the number of active TCP connections of each IPv6 node. The determination of the IPv6 path hop count is achieved by executing the Traceroute6 tool to trace the path to the target address and statistically calculating the number of three-layer network device forwarding times between the client subnet and the target server. For example, if the measured path is client gateway → core switch → border router → IDC access switch → target server, the hop count is recorded as 4.

[0035] Step 1033: Compare the first set of real-time parameters with the preset IPv4 reference parameters to generate an IPv4 link quality score; compare the second set of real-time parameters with the preset IPv6 reference parameters to generate an IPv6 link quality score.

[0036] In the embodiment of this application, the quality assessment module of the dual-stack network adaptive scheduling system compares the first set of real-time parameters with the preset IPv4 reference parameters to generate an IPv4 link quality score. The above comparison process adopts the multi-dimensional parameter normalization algorithm. For example, when the current IPv4 link delay is 28 ms (the reference value is 30 ms), the available bandwidth is 95 Mbps (the reference value is 100 Mbps), and the request response success rate is 99% (the reference value is 98%), the dual-stack network adaptive scheduling system obtains the comprehensive score through the weighted calculation formula. Exemplarily, the weight of the delay parameter is set to 40%, the weight of the bandwidth is 35%, and the weight of the success rate is 25%, and finally a score result 3% higher than the reference value is generated. In the synchronous calculation of the IPv6 link quality score, the system compares the throughput, load balancing metric, and path hop count in the second set of real-time parameters with the IPv6 performance baseline set by the enterprise network operation and maintenance department respectively. For example, if the measured IPv6 throughput reaches 110% of the reference value, the load metric is 15% better than the baseline, and the path hop count is 1 more than expected, the final score is obtained by dynamically adjusting the weight coefficients of each parameter.

[0037] Step 1034: Dynamically adjust the sorting weights of IPv4 and IPv6 in the address type priority list according to the difference between the IPv4 link quality score and the IPv6 link quality score.

[0038] In the embodiment of the present application, the weight adjustment module of the dual-stack network adaptive scheduling system dynamically adjusts the sorting weights of the address type priority list according to the difference between the IPv4 and IPv6 link quality scores. When it is monitored that the IPv4 link quality score of the domain name "file.example.com" is 85 points, the IPv6 score is 78 points, and the difference exceeds the preset first threshold of 10 points, the dual-stack network adaptive scheduling system will raise the sorting level of IPv4 in the priority list by two levels.

[0039] On the contrary, if the IPv6 score of the same domain name reaches 92 points during the business peak period while the IPv4 score drops to 80 points and the difference exceeds the second threshold of 8 points, the IPv6 priority policy upgrade is triggered. When the score difference between the two is in the dynamic adjustment range of 5-7 points, the system maintains the existing weight configuration unchanged to ensure the stability of the scheduling strategy.

[0040] At the same time, the dual-stack network adaptive scheduling system retrieves the historical connection records of this domain name in the past 24 hours and finds that the successful connection rate of the IPv6 address continuously remains above 99.5%, while there is a 2% failure rate for the IPv4 address. Based on this, a +5% success rate compensation correction is made to the weight generated by the real-time score.

[0041] Step 1035: Generate an adaptive scheduling strategy including the dynamic shunt ratio of IPv4 addresses and IPv6 addresses based on the adjusted sorting weights, and the dynamic shunt ratio is used to control the address allocation priorities of different protocol stacks.

[0042] In the embodiment of the present application, the policy generation module of the dual-stack network adaptive scheduling system generates an adaptive scheduling strategy including the dynamic shunt ratio of IPv4 and IPv6 addresses based on the adjusted sorting weights. For the domain name of the "mail.example.com" mail system, when the real-time weight shows that the IPv6 priority is 65% and the IPv4 is 35%, the dual-stack network adaptive scheduling system will be configured to return a mixed parsing result of 3 IPv6 addresses and 2 IPv4 addresses to the dual-protocol stack client, forming an address allocation ratio of 6:4. This address allocation ratio is achieved through the sorting of multiple resource records in the DNS response message. The AAAA record of the IPv6 address is arranged before the A record to guide most clients to establish IPv6 connections first. At the same time, the dual-stack network adaptive scheduling system deploys a protocol type counter at the enterprise wide area network export to monitor the actual usage rate of different address types in real time. When it is detected that the IPv6 connection ratio is lower than the preset target value, the weight readjustment process is automatically triggered.

[0043] In the next step, dynamically adjust the sorting weights of IPv4 and IPv6 in the address type priority list according to the difference between the IPv4 link quality score and the IPv6 link quality score in step 1034, including:

[0044] (1) If the IPv4 link quality score is higher than the IPv6 link quality score and the difference exceeds the first threshold, then increase the sorting level of IPv4 in the address type priority list.

[0045] In this step, when the dual-stack network adaptive scheduling system determines that the IPv4 link quality score is higher than that of IPv6 and the difference exceeds the first threshold, perform the protocol stack priority adjustment operation. For example, for the video conferencing system domain name "meet.example.com", when the IPv4 link latency is stable at 25 ms and the IPv6 link latency reaches 45 ms due to incomplete path optimization, the dual-stack network adaptive scheduling system compares the quality score difference between the two to be 15 points (exceeding the set threshold of 10 points), and immediately raises the sorting level of IPv4 in the address type priority list from the second level to the first level. This adjustment is achieved by modifying the protocol stack priority field in the policy database, so that the IPv4 address is preferentially selected during subsequent parsing request processing. Further, the promotion operation follows the principle of progressive adjustment, with each level change not exceeding two gradients to prevent policy oscillations caused by instantaneous fluctuations in the network state.

[0046] (2) If the IPv6 link quality score is higher than the IPv4 link quality score and the difference exceeds the second threshold, then increase the sorting level of IPv6.

[0047] In this step, when the IPv6 link quality score is significantly better than that of IPv4, the dual-stack network adaptive scheduling system starts the IPv6 priority enhancement mechanism. Taking the file transfer service domain name "ftp.example.com" as an example, when its IPv6 path reaches the throughput of 1.2 Gbps and the number of hops is reduced to 3 hops through the newly deployed dedicated backbone network direct link, while the IPv4 link has a throughput of only 800 Mbps due to limited cross-operator interconnection bandwidth, the dual-stack network adaptive scheduling system calculates that the quality score difference between the two reaches 12 points (exceeding the second threshold of 8 points), and thus directly raises the sorting level of IPv6 from the default third level to the highest level. After the adjustment takes effect, all IPv6-enabled clients will first obtain the AAAA record when resolving this domain name, and only try the IPv4 address when the IPv6 connection fails. Then, the dual-stack network adaptive scheduling system synchronously updates the preferred path mark in the network topology map to provide a reference basis for subsequent service quality evaluation.

[0048] (3)When the difference in link quality scores between IPv4 and IPv6 is between the first threshold and the second threshold, keep the current sorting weight unchanged.

[0049] In this step, the dual-stack network adaptive scheduling system maintains the current sorting weight when the network state fluctuates slightly. For example, for the domain name "oa.example.com" of the office OA system, the difference in link quality scores between IPv4 and IPv6 during weekdays continuously fluctuates within the range of 5 - 7 points and does not reach the adjustment threshold of 8 points. The dual-stack network adaptive scheduling system keeps the existing IPv6 priority level unchanged. This stability guarantee mechanism is achieved by setting a status maintenance timer. When the difference does not exceed the threshold in the last three evaluation cycles, the weight parameter is locked for at least 30 minutes. During this period, the dual-stack network adaptive scheduling system continuously collects new network performance data but does not trigger weight adjustment, avoiding frequent policy changes caused by short-term network congestion and ensuring the continuity of client connections.

[0050] (4)Compensate and correct the sorting weight according to the successful connection rates of IPv4 and IPv6 in the historical scheduling records.

[0051] In this step, the historical data analysis module of the dual-stack network adaptive scheduling system compensates and corrects the sorting weight. For the domain name "crm.example.com" of the CRM system, although real-time monitoring shows that the IPv6 link quality score is slightly higher than that of IPv4, by retrieving the connection records in the past 7 days, it is found that the success rate of IPv6 addresses is only 92%, while the IPv4 success rate remains at 98%. The system calculates that the difference in historical success rates between the two reaches 6 percentage points, exceeding the preset compensation threshold of 5%. Therefore, a compensation coefficient of -10% is applied to the weight generated by the current real-time score. This compensation mechanism is achieved by introducing a time decay factor, with the weight of recent historical data accounting for 70% and that of long-term data accounting for 30%, ensuring that the correction result reflects both the long-term trend and the latest network state.

[0052] (5)Perform weighted fusion on the compensated and corrected sorting weight and the real-time link quality score to generate the final sorting weight to update the address type priority list.

[0053] In this step, the dual-stack network adaptive scheduling system generates the final sorting weight through a weighted fusion algorithm. Taking the enterprise portal domain name "www.example.com" as an example, the difference in the real-time link quality score shows that IPv6 is 9 points better than IPv4, so the weight of IPv6 should be increased. At the same time, the historical success rate analysis shows that the IPv6 connection success rate is 3 percentage points higher than that of IPv4, resulting in a compensation coefficient of +5%. For example, the two can be combined in a linear superposition manner, where the real-time score accounts for 70% of the weight and the historical compensation accounts for 30%. Finally, a comprehensive decision value with a 12% increase in the priority of IPv6 is generated. This fusion result is written into the dynamic policy library and synchronized to all DNS resolution nodes through distributed caching to ensure the consistency of the network-wide scheduling policy.

[0054] In the next step, the sorting weight is compensated and corrected according to the successful connection rates of IPv4 and IPv6 in the historical scheduling records, including:

[0055] Step 10341: Extract the ratio of the cumulative number of successful connections to the number of failed connections of IPv4 addresses under the target domain name within a preset historical time period as the IPv4 historical success rate.

[0056] In Step 10341, the log analysis unit of the dual-stack network adaptive scheduling system extracts the IPv4 connection data within a preset historical time period from the connection tracking database. For the VPN access domain name "vpn.example.com", among the 12,000 connections established by the client using IPv4 addresses in the recent 24 hours, 11,760 times successfully completed the TCP handshake, and the failure records included 240 timeouts and 20 connection rejections. By calculating the ratio of the number of successful connections to the number of failed connections, the IPv4 historical success rate is obtained as 11760 / (11760 + 260) = 97.8%. This process uses a sliding time window mechanism to continuously update the statistical results at a granularity of 1 hour to ensure the timeliness and accuracy of historical data.

[0057] Step 10342: Extract the ratio of the cumulative number of successful connections to the number of failed connections of IPv6 addresses within the same time period as the IPv6 historical success rate.

[0058] In step 10342, continuing with "vpn.example.com" as an example, 8500 IPv6 connection attempts were monitored within the same 24 hours, of which 8320 were successful. The reasons for failure included 150 cases of network unreachability and 30 cases of certificate errors. By excluding application layer errors (such as certificate issues) and only counting the number of network layer failures, the historical IPv6 success rate was calculated as 8320 / (8320 + 150) = 98.2%. The system uses independent data processing threads to process IPv4 and IPv6 logs respectively, and ensures data consistency during concurrent access through atomic counters.

[0059] Step 10343: Calculate the ratio difference between the historical IPv4 success rate and the historical IPv6 success rate; if the ratio difference exceeds a preset compensation threshold, increase the current sorting weight by the difference compensation coefficient.

[0060] In step 10343, the difference calculation module of the dual-stack network adaptive scheduling system can compare the historical success rates of IPv4 and IPv6. Taking the aforementioned "vpn.example.com" as an example, compared with the IPv4 success rate of 97.8%, the absolute difference in the IPv6 success rate of 98.2% is 0.4 percentage points, and the relative difference ratio is 1.004. When this ratio exceeds the preset compensation threshold of 1.005, the compensation mechanism is triggered. The calculation process uses double-precision floating-point arithmetic to retain four decimal places of precision, and excludes the influence of occasional network jitters through an outlier filtering algorithm, for example, ignoring local failure events with a duration of less than 5 minutes.

[0061] Step 10344: Linearly superimpose the difference compensation coefficient and the real-time link quality score to generate a compensated sorting weight.

[0062] In step 10344, the difference compensation coefficient and the real-time link quality score are linearly superimposed. Taking the enterprise cloud disk domain name "cloud.example.com" as an example, the difference in its real-time score showing that the IPv6 quality is better than IPv4 is 7 points, which should result in a +5% adjustment of the basic weight. At the same time, the historical success rate analysis shows that IPv6 is 1.2 percentage points higher than IPv4, triggering a +3% compensation coefficient. The final adjustment value of +4.4% is calculated through the weighted formula (real-time score × 0.7 + historical compensation × 0.3), and this value is converted into a specific sorting change in the priority list. The superimposition process follows the saturation operation rule, limiting the maximum amplitude of a single adjustment to no more than 15% to prevent policy mutations caused by extreme network conditions.

[0063] It can be understood that the above-mentioned adaptive scheduling policy generation process realizes dynamic optimization through a multi-layer feedback mechanism, taking into account historical experience while ensuring real-time performance. The above embodiments deploy each functional module using a microservices architecture and perform asynchronous communication through a message queue to ensure a millisecond-level response speed even under high-concurrency parsing requests. Further, after the policy takes effect, key indicators such as the client connection establishment duration and data transmission rate are continuously monitored to form a closed-loop optimization system, promoting the steady development of the dual-stack evolution of the enterprise network towards the established service quality goals.

[0064] In one implementation, the selecting of the target IP address type from the pre-configured IP address pool based on the adaptive scheduling policy in step 104 includes:

[0065] Step 1041: When the adaptive scheduling policy indicates to adopt the protocol stack parallel mode, extract the IPv4 address list and the IPv6 address list from the IP address pool simultaneously.

[0066] In this embodiment, the address selection module of the dual-stack network adaptive scheduling system performs operations in the protocol stack parallel mode in the enterprise office network environment. When the system determines that the business system corresponding to the target domain name "oa.example.com" has both IPv4 and IPv6 service capabilities and the client supports the dual protocol stack, it extracts the IPv4 address list and the IPv6 address list from the pre-configured IP address pool in parallel. This IP address pool is synchronized through the data interface with the enterprise network infrastructure management platform, where the IPv4 address list contains two A records of the preset 192.0.2.45 and 192.0.2.46, and the IPv6 address list contains two AAAA records of 2001:db8::1:10 and 2001:db8::1:11. Through the address validity verification mechanism, preprocessing is performed on the two lists to exclude IP addresses in the maintenance state or those that fail the health check, ensuring that each entry in the candidate address set has routability and service availability.

[0067] Step 1042: Determine the primary and secondary allocation orders of the IPv4 and IPv6 addresses according to the request protocol type of the client.

[0068] In this embodiment, the protocol decision module of the dual-stack network adaptive scheduling system determines the address allocation order according to the protocol type of the client request. For clients accessing through the enterprise wireless network, the EDNSClient Subnet extension field in the DNS request packet indicates that the client supports both IPv4 and IPv6 protocol stacks. The dual-stack network adaptive scheduling system sets the IPv6 address as the primary allocation order and the IPv4 address as the secondary allocation order with reference to the preset address type priority policy. This decision-making process is dynamically adjusted in combination with real-time network status data. For example, when it is detected that there is regional congestion in the IPv6 backbone link, the dual-stack network adaptive scheduling system temporarily sets the IPv4 as the primary allocation order. For clients that only support the IPv4 protocol, the dual-stack network adaptive scheduling system automatically sets the IPv4 address list as the primary allocation order and triggers a protocol conversion service alarm when there is no available IPv4 address.

[0069] Step 1043: Encapsulate the IPv4 or IPv6 address corresponding to the primary allocation order into the preferred record field of the DNS response packet.

[0070] In this embodiment, the response generation module of the dual-stack network adaptive scheduling system encapsulates the address corresponding to the primary allocation order into the preferred record field of the DNS response packet. Taking the enterprise file service domain name "file.example.com" as an example, when the primary allocation order is set to IPv6 priority, the dual-stack network adaptive scheduling system arranges the AAAA records of the IPv6 addresses 2001:db8::1:10 and 2001:db8::1:11 in the first two positions of the DNS packet Answer Section. This encapsulation process follows the resource record sorting rules specified in RFC 1035 to ensure that the client preferentially attempts the addresses arranged at the front when parsing the response. The dual-stack network adaptive scheduling system also performs address geographical location matching calculations. When it is detected that the request source belongs to the subnet of an enterprise branch, it preferentially returns the IPv6 address entry with the closest network topology distance to this area.

[0071] Step 1044: Add the secondary protocol stack address corresponding to the secondary allocation order to the additional record field of the DNS response packet.

[0072] In this embodiment, the standby address processing unit of the dual-stack network adaptive scheduling system adds the standby protocol stack address to the additional record field of the DNS response message. For the resolution request of the video conferencing system domain name "meet.example.com", after the dual-stack network adaptive scheduling system fills in the IPv6 address 2001:db8::1:20 in the main record field of the Answer Section, it fills in the A record of the additional IPv4 address 192.0.2.55 in the Additional Section. This additional record adopts the standard resource record format and contains the same domain name, record type, and time-to-live value as the main record. The dual-stack network adaptive scheduling system distinguishes the primary and standby address types by setting the protocol stack identification bit. When the client supports the protocol stack fallback mechanism, fast failover can be achieved using the additional record.

[0073] Step 1045: In response to the failure of the client to connect to the primary address, trigger a protocol stack switching process based on the standby address in the additional record field to guide the client to re-initiate a connection request.

[0074] In this embodiment, the connection monitoring module of the dual-stack network adaptive scheduling system responds to the event of the client's failure to connect to the primary address and starts the protocol stack switching process. When the enterprise mobile office client attempts to access the "oa.example.com" service through the IPv6 address 2001:db8::1:10, if the TCP three-way handshake is not completed within the preset timeout of 3000 milliseconds, the dual-stack network adaptive scheduling system captures this connection exception event through the deep packet inspection device in the enterprise network. The protocol stack switching instruction generation unit then triggers the address fallback mechanism, resolves the IPv4 standby address 192.0.2.45 in the additional record field of the DNS response message, and sends an ICMP redirect message to the client through the policy execution node in the enterprise intranet. This message contains the protocol stack switching identifier and the standby address information.

[0075] Among them, the protocol stack switching process includes:

[0076] (1) Monitor the connection establishment status between the client and the primary address. If no connection confirmation signal is received within the preset timeout, generate a protocol stack switching instruction.

[0077] Based on this step, the connection status monitoring unit of the dual-stack network adaptive scheduling system tracks the connection establishment process between the client and the primary address in real time. For the scenario where the enterprise R & D department client accesses the code repository domain name "git.example.com", the TCP SYN packet sending event is captured through a kernel-level network probe, and the connection establishment countdown timer is started. When SYN retransmission fails three times in a row without receiving a SYN-ACK response, and the cumulative time exceeds the 2000-millisecond threshold, a protocol stack switching instruction containing the failure reason code is generated. This monitoring mechanism is synchronized with the session tracking table of the enterprise firewall and can accurately distinguish between network layer connection failures and application layer service unavailability states.

[0078] (2)Parse the additional record field of the DNS response packet and extract the protocol stack type and IP address corresponding to the alternate address.

[0079] Based on this step, the packet parsing module of the dual-stack network adaptive scheduling system deeply decodes the additional record field of the DNS response packet. Taking the resolution response of the enterprise mail system domain name "mail.example.com" as an example, the A record of the alternate IPv4 address 192.0.2.60 is extracted from the Additional Section, and the protocol stack type identifier corresponding to this address is obtained at the same time. The parsing process strictly follows the DNS packet format specification and processes operations such as resource record name compression, TTL value parsing, and RDATA field decoding. By establishing an address type mapping table, the compatibility verification between the alternate address and the client protocol stack capabilities is ensured.

[0080] (3)Send an address redirection packet to the client, and the address redirection packet contains the alternate address and the protocol stack switching identifier.

[0081] Based on this step, the redirection control module of the dual-stack network adaptive scheduling system sends an address redirection packet to the client. When the market department client fails to access the primary IPv6 address of "crm.example.com", an ICMPv6 Destination Unreachable packet containing the protocol stack switching identifier 0xA6 is generated, and the alternate IPv4 address 192.0.2.70 is embedded in its payload part. This redirection packet is sent to the requesting source client through the policy routing function of the enterprise core switch, while retaining the port number and session identification information of the original request. By setting the priority of the DSCP field of the IP packet, it is ensured that the redirection instruction can still be delivered in a timely manner under network congestion conditions.

[0082] (4)Record the protocol stack type, switching timestamp, and switching reason of this switching operation and update them to the historical switching log.

[0083] Based on this step, the logging unit of the dual-stack network adaptive scheduling system writes the details of the switching operation into the historical switching log. For the protocol stack switching event of the financial system domain name "erp.example.com", metadata such as the switching timestamp (accurate to milliseconds), the original primary IPv6 address 2001:db8::1:30, the target standby IPv4 address 192.0.2.75, and the failure reason code (network unreachable) are recorded. These log entries are transmitted to the central analysis platform through an encrypted channel and correlated with the alarm events of the enterprise network performance monitoring system to form a complete fault diagnosis evidence chain.

[0084] (5)Dynamically adjust the allocation ratio of the primary and standby addresses in the subsequent scheduling policy according to the cumulative number of switches of the same target domain name in the historical switching log.

[0085] Based on this step, the policy optimization module of the dual-stack network adaptive scheduling system dynamically adjusts the allocation ratio of the primary and standby addresses according to the historical switching log. When it is analyzed that the number of times the "file.example.com" domain name has switched from the primary IPv6 address to the standby IPv4 address in the past 24 hours exceeds 50 times, the IPv6 allocation weight of the primary address is automatically reduced by 20%, and at the same time, the priority of the IPv4 address in the standby allocation order is increased. This adjustment process uses a sliding time window algorithm, focusing on the changing trend of the switching frequency in the most recent 2 hours to ensure that the scheduling policy responds in a timely manner to network state fluctuations. The optimized allocation ratio is synchronized to all DNS resolution nodes in real time through the configuration management interface, forming a closed-loop control mechanism.

[0086] Applying the above embodiments, the entire address selection and protocol stack switching process realizes intelligent scheduling through a multi-layer collaborative mechanism, optimizing the utilization rate of network resources while ensuring service continuity. By deploying monitoring agents at key nodes of the enterprise network, an end-to-end connection quality perception system is constructed, enabling the protocol stack switching decision to not only conform to the global scheduling policy but also adapt to the dynamic changes of the local network environment. It can be seen that the above embodiments effectively solve the problem of rigid protocol switching in traditional dual-stack networks, significantly improving the reliability and user experience of enterprise employees accessing critical business systems.

[0087] In another implementation, the method further includes: deploying a protocol stack compatibility detection service at the DNS resolution layer. Among them, the following operations are performed by the protocol stack compatibility detection service:

[0088] Step 201: Send an IPv4 protocol detection message and an IPv6 protocol detection message to the client, and both the IPv4 protocol detection message and the IPv6 protocol detection message contain a preset protocol identifier.

[0089] In step 201, the protocol detection module of the dual-stack network adaptive scheduling system sends IPv4 protocol detection messages and IPv6 protocol detection messages to the client through the core switch of the enterprise office network. The IPv4 protocol detection message is encapsulated using the UDP protocol, and the destination port is set to the standard DNS service port 53. A preset protocol identifier "IPv4_Probe_OA_v1.2" is embedded in the application layer payload. The IPv6 protocol detection message is in the ICMPv6 protocol format, with its type field set to an information request message, and a unique identifier "IPv6_Probe_OA_v1.2" is carried in the option field.

[0090] Taking the access of the enterprise human resources system client to the domain name "hr.example.com" as an example, when the client first accesses the enterprise wireless network, its ARP request message is captured through the traffic mirroring port of the core switch, triggering the protocol stack detection process. After the detection message is filtered by the security policy of the enterprise firewall, the edge router adds a priority mark and forwards it to the target client.

[0091] Step 202: Listen for the feedback behavior of the client to the IPv4 protocol detection message and the IPv6 protocol detection message within a preset response time.

[0092] In step 202, the response monitoring unit of the dual-stack network adaptive scheduling system monitors the client feedback behavior within a preset response time window. For the detection request of the enterprise conference room intelligent terminal device, the global response timeout threshold is set to 3000 milliseconds, and the response messages from the client are captured in real time through the flow analysis probe deployed on the core switch. When the client network interface card supports the IPv6 protocol stack, it will return an ICMPv6 reply message; if it only supports the IPv4 protocol stack, it will respond with a TCP RST message or remain silent.

[0093] Taking the access of the R & D department workstation to the code management platform "git.example.com" as an example, after the dual-stack network adaptive scheduling system sends the dual-protocol detection message, it analyzes the protocol type field of the returned message through deep packet inspection technology, records the accurate response timestamp and protocol interaction characteristics. The session state table configured by the enterprise firewall is updated synchronously to ensure that the detection response message will not be misintercepted by the security policy.

[0094] Step 203: If only the IPv4 detection response is received, mark the protocol stack support status of the client as only IPv4 compatible; if only the IPv6 detection response is received, mark the protocol stack support status of the client as only IPv6 compatible; if both the IPv4 detection response and the IPv6 detection response are received, mark the protocol stack support status of the client as dual-stack compatible.

[0095] In step 203, the status marking module of the dual-stack network adaptive scheduling system updates the client protocol stack support status according to the characteristics of the response message. When the mobile office device of the marketing department only returns an IPv4 protocol detection response, the dual-stack network adaptive scheduling system sets the status corresponding to the MAC address of this client in the protocol stack support status table to be IPv4-only compatible.

[0096] For newly deployed IPv6 single-stack IoT devices, after the dual-stack network adaptive scheduling system detects that they only respond to IPv6 protocol detection messages, they are marked as IPv6-only compatible status. Taking the 4K smart terminal in the enterprise video conference room as an example, this device returns both IPv4 and IPv6 detection response messages. The dual-stack network adaptive scheduling system confirms its dual-stack compatibility ability by parsing the protocol identifier matching field in the response message and sets the dual-protocol stack support flag in the status table. This marking process is cross-validated in combination with the OUI (Organizationally Unique Identifier) information of the client network interface card to improve the accuracy of status determination.

[0097] Step 204: Store the marking result in the protocol stack support status table, and the protocol stack support status table is the direct call object for subsequent DNS resolution requests.

[0098] In step 204, the data storage module of the dual-stack network adaptive scheduling system persistently stores the protocol stack support status marking result in a distributed database. The protocol stack support status table adopts a three-copy storage mechanism and includes three core fields: the client MAC address, the most recent detection timestamp, and the protocol stack compatibility status. Taking the encrypted terminal device of the finance department as an example, its dual-stack compatibility status marking record includes the MAC address 00:1A:3F:2B:55:CC, the most recent update time 2023-08-20T14:30:00Z, and the status value Dual-Stack. This table synchronizes data with the employee terminal device management system through the enterprise intranet LDAP service to ensure that the protocol stack information of newly added devices can be included in a timely manner. The dual-stack network adaptive scheduling system establishes an hourly incremental backup mechanism to prevent misjudgment of status caused by data loss.

[0099] Furthermore, the calling process of the protocol stack support status table includes:

[0100] Step 301: When receiving a new DNS resolution request, extract the source IP address and request protocol type of the new client corresponding to the new DNS resolution request.

[0101] In step 301, when processing a new DNS resolution request, the request parsing unit of the dual-stack network adaptive scheduling system performs source information extraction operations. When an enterprise mobile office employee accesses "oa.example.com" through a VPN connection, the dual-stack network adaptive scheduling system extracts the source IPv4 address 203.0.113.25 from the IP header of the DNS request message and simultaneously parses the request protocol type identifier in the EDNS extension field. This extraction process uses zero-copy technology to directly access the network interface card receive buffer to ensure processing efficiency in high-concurrency scenarios. For clients accessing through the IPv6 protocol, the dual-stack network adaptive scheduling system also records its IPv6 source address 2001:db8:cafe::1a3f:2b55:cc and establishes a dual-protocol address association mapping table.

[0102] Step 302: Query the historical marking result associated with the source IP address in the protocol stack support status table.

[0103] In step 302, the status query module of the dual-stack network adaptive scheduling system retrieves the protocol stack support status table by the source IP address. Taking the example of a branch office client accessing the enterprise file server "file.example.com", the dual-stack network adaptive scheduling system queries the distributed database according to the source IPv4 address 192.168.12.34 and returns the historical marking result of this client as dual-stack compatible. The query process uses a Bloom filter for pre-screening to quickly exclude client devices without historical records. For enterprise visitor network clients using NAT conversion, the dual-stack network adaptive scheduling system associates the internal device list through the mapped public IP address to ensure accurate traceability of the protocol stack status information.

[0104] Step 303: If the historical marking result indicates that the new client is dual-stack compatible, allocate an IPv4 address or an IPv6 address according to the current adaptive scheduling policy.

[0105] In step 303, the policy execution module of the dual-stack network adaptive scheduling system implements a dynamic address allocation policy for dual-stack compatible clients. When it detects that a design department workstation accesses "cad.example.com", the dual-stack network adaptive scheduling system refers to the current network status assessment result and preferentially allocates the IPv6 address 2001:db8::cad:1. This allocation policy is achieved through the sorting of AAAA records in the DNS response message, placing the preferred address at the top of the resource record list. The dual-stack network adaptive scheduling system also monitors the real-time quality metrics of the IPv6 link. If abnormal traffic is found on the IPv6 port of the target server, it automatically switches to the IPv4 address allocation mode. The policy execution process is linked with the enterprise software-defined network controller to dynamically adjust the bandwidth allocation ratio of each protocol stack.

[0106] Step 304: If the historical marking result is single protocol stack compatibility, skip the adaptive scheduling policy generation step and directly allocate the corresponding protocol stack address.

[0107] In step 304, the direct allocation module of the dual-stack network adaptive scheduling system implements a fast response mechanism for single protocol stack clients. For the scenario where an old version attendance terminal that only supports the IPv4 protocol accesses "attendance.example.com", the dual-stack network adaptive scheduling system skips the complex policy calculation process and directly returns the A record of the IPv4 address 192.0.2.101. This process is implemented through a pre-generated protocol stack address mapping table, and a direct cache channel is established for known single protocol clients. The dual-stack network adaptive scheduling system also records the log information of the direct allocation operation for the subsequent device compatibility analysis of the network transformation project.

[0108] Step 305: When the protocol stack support status of the new client changes, update the protocol stack support status table through periodic re-probing.

[0109] In step 305, the status maintenance service of the dual-stack network adaptive scheduling system updates the protocol stack support status table through a periodic re-probing mechanism. For the smart whiteboard device that upgrades the network interface card in the enterprise conference room, the dual-stack network adaptive scheduling system sends dual-protocol probing messages every 24 hours. When detecting the new IPv6 protocol response capability, the status flag is updated to dual-stack compatible. The re-probing process adopts a progressive time interval algorithm, shortening the probing period to 6 hours for clients with frequently changing protocols and extending it to 72 hours for devices with stable status. The update operation ensures data consistency through a two-phase commit protocol to avoid incorrect updates caused by network jitter. In addition, the dual-stack network adaptive scheduling system synchronously clears the client records that have been inactive for more than 90 days to maintain the high query performance of the status table.

[0110] Applying the above embodiments, the protocol stack compatibility detection and status management mechanism works collaboratively through multiple layers to build an accurate client capability profile. While ensuring the DNS resolution efficiency, it provides a reliable decision-making basis for the adaptive scheduling policy, effectively improving the service quality and operation and maintenance transparency of the enterprise network during the dual-stack evolution process. It can be understood that through the detection proxy cluster deployed on the core network nodes, the protocol stack status of all network clients can be perceived in real time, and combined with historical data analysis to predict the device upgrade trend, providing data support for the enterprise to formulate network transformation plans.

[0111] In a preferred implementation, the "When the protocol stack support status of the new client changes, update the protocol stack support status table through periodic re-probing" in step 305 includes:

[0112] Step 30501: Deploy a timer trigger mechanism at the DNS resolution layer and start a periodic detection timer associated with the protocol stack support status table.

[0113] In this embodiment, the timer management module of the dual-stack network adaptive scheduling system deploys a timer trigger mechanism through the core DNS server of the enterprise office network. When the system detects that there are more than 1000 client records in the protocol stack support status table, it automatically starts the periodic detection timer service. This timer service keeps clock synchronization with the enterprise network time protocol server and adopts a hierarchical trigger strategy: the core network device triggers the main timer every 5 minutes, and the edge access layer devices start the sub-timers synchronously according to regions. Taking the enterprise headquarters data center as an example, when the dual-stack network adaptive scheduling system performs a full-scale client status scan at 3:00 am every day, it synchronously initializes the periodic detection timer cluster to ensure load balancing of detection tasks in different business regions.

[0114] Step 30502: When the periodic detection timer reaches the preset re-detection period, scan all the protocol stack support status entries of the clients in the protocol stack support status table.

[0115] In this embodiment, after the periodic detection timer is triggered, the status scan module of the dual-stack network adaptive scheduling system starts the full-scale scan process of the protocol stack support status table. When the re-detection period of the enterprise branch network reaches the preset 24-hour threshold, the dual-stack network adaptive scheduling system reads the client status entries in batches from the distributed database. The scan process uses cursor paging technology to process 500 records each time to avoid memory overflow. Taking the client devices of the enterprise R & D department as an example, the dual-stack network adaptive scheduling system traverses the status table copies stored in three availability zones, compares the data consistency between the copies, and generates a pending processing queue.

[0116] Step 30503: Extract the last detection timestamp and the protocol stack support status flag value corresponding to the source IP address of each client.

[0117] In this embodiment, the metadata extraction component of the dual-stack network adaptive scheduling system extracts key time parameters from the protocol stack support status table. For the records of the encrypted terminals in the finance department, the dual-stack network adaptive scheduling system extracts the last detection timestamp field value "2023-08-20T14:30:00Z" and the protocol stack support status flag value "Dual-Stack". This extraction process uses columnar storage optimization technology to only read the necessary fields to improve processing efficiency. The dual-stack network adaptive scheduling system also verifies the legality of the timestamp and filters out abnormal records with future timestamps caused by network latency.

[0118] Step 30504: Calculate the time interval between the current system time and the last detection timestamp, and compare the difference with the preset re-detection period.

[0119] In this embodiment, the timeliness calculation module of the dual-stack network adaptive scheduling system performs the time interval comparison operation. Taking the last detection time of the mobile office device in the marketing department, "2023-08-20T10:15:00" as an example, after the dual-stack network adaptive scheduling system obtains the current Coordinated Universal Time "2023-08-21T09:00:00", the calculated time interval is 22 hours and 45 minutes. This interval value is compared with the preset re-detection period of 24 hours to generate an evaluation result of the remaining effective time of 1 hour and 15 minutes. The above calculation process uses a sliding time window algorithm to dynamically adapt to client devices in different time zones.

[0120] Step 30505: If the time interval exceeds the validity threshold of the preset re-detection period, add the source IP address of the client to the re-detection queue.

[0121] In this embodiment, the queue generation module of the dual-stack network adaptive scheduling system screens the devices to be detected according to the timeliness evaluation results. When it is detected that the last detection time of the terminal device in the human resources department has exceeded the 72-hour validity threshold, the dual-stack network adaptive scheduling system adds its IPv4 address 192.168.5.20 to the re-detection queue. Among them, the queue generation process adopts a priority sorting strategy, arranging the client devices of the enterprise's key business systems (such as the access terminals of the ERP system) at the front of the queue to ensure priority re-detection. In addition, the queue management module simultaneously implements duplicate removal verification to prevent repeated enqueueing caused by multiple IP bindings of the same device.

[0122] Step 30506: Send IPv4 protocol probe packets and IPv6 protocol probe packets to each client according to the order of the source IP addresses in the re-detection queue.

[0123] In this embodiment, the probe packet distribution unit of the dual-stack network adaptive scheduling system initiates protocol detection according to the queue order. For example, for the 50th device in the re-detection queue: the intelligent display screen in the conference room (IP address 203.0.113.88), the dual-stack network adaptive scheduling system sends a UDP probe packet carrying the identifier "IPv4_Probe_V2.1" to port 53 through the policy routing port of the enterprise core switch, and at the same time sends an ICMPv6 information request packet containing the "IPv6_Probe_V2.1" identifier. Specifically, the packet sending interval is set to 500 milliseconds to avoid network congestion, and each client sends three probe packets continuously to ensure reliability.

[0124] Step 30507: Listen for the type of the probe response message returned by the client within a preset response time window.

[0125] In this embodiment, the response listening service of the dual-stack network adaptive scheduling system captures the returned message within a preset time window of 3000 milliseconds. For the probe request of the enterprise warehouse management terminal, the dual-stack network adaptive scheduling system collects response data through the traffic mirroring port deployed on the access switch. When the terminal only returns an IPv4 protocol probe response, the listening service records its MAC address 00:1A:3F:2B:55:CD and the response timestamp. Among them, the response message type recognition module analyzes the transport layer protocol characteristics at the same time to distinguish the TCP RST message of IPv4 from the ICMPv6 reply message of IPv6.

[0126] Step 30508: Extract the protocol identifier and the client source IP address carried in the probe response message.

[0127] In this embodiment, the message parsing component of the dual-stack network adaptive scheduling system extracts the protocol characteristics in the probe response. Taking the response message of the design department workstation as an example, the system decodes the protocol identifier "IPv6_Probe_V2.1" from the extension header of the IPv6 probe reply message and extracts the source IPv6 address 2001:db8:cafe::1a3f:2b55:cc. Strict security checks are implemented during the parsing process, including verifying the hash signature of the identifier and checking the message integrity check value to prevent probe response attacks.

[0128] Step 30509: If the protocol identifier indicates that the client returns an IPv4 probe response, update the IPv4 compatibility flag corresponding to the source IP address in the protocol stack support status table to the valid state; if the protocol identifier indicates that the client returns an IPv6 probe response, update the IPv6 compatibility flag corresponding to the source IP address in the protocol stack support status table to the valid state; if no IPv4 probe response or IPv6 probe response is received within the response time window, set the corresponding protocol stack compatibility flag to the invalid state.

[0129] In this embodiment, the status update module of the dual-stack network adaptive scheduling system refreshes the protocol stack support status table according to the response result. When the enterprise video conferencing terminal returns dual-protocol probe responses of IPv4 and IPv6, the dual-stack network adaptive scheduling system sets both its IPv4 compatibility flag and IPv6 compatibility flag to the valid state. For the old version of the attendance machine that does not respond to all three probes, the dual-stack network adaptive scheduling system sets its IPv4 compatibility flag to the invalid state and triggers a device offline alarm. The update operation of this step adopts atomic transaction processing to ensure the ultimate consistency of each replica in the distributed database cluster.

[0130] Step 30510: Recalculate the protocol stack support status flag value of the client according to the updated protocol stack compatibility flag; overwrite the historical flag value in the protocol stack support status table with the recalculated protocol stack support status flag value.

[0131] In this embodiment, the flag calculation module of the dual-stack network adaptive scheduling system re-evaluates the protocol support capabilities of the client. After re-probing, a certain mobile office device only returns an IPv6 probe response. The dual-stack network adaptive scheduling system changes the protocol stack support status flag value from "Dual-Stack" to "IPv6-Only" according to the latest result. The calculation process of this step adopts a state machine model, defining the mapping relationship between valid flag combinations and final states. For example, when the IPv4 flag fails and the IPv6 flag is valid, a new state code "0x6A" is generated.

[0132] Step 30511: Record the timestamp of this re-probing operation to the last probe timestamp field in the protocol stack support status table.

[0133] In this embodiment, the timestamp recording service of the dual-stack network adaptive scheduling system updates the timeliness information of the protocol stack support status table. After completing the re-probing of the R & D department's test terminal (IP address 192.168.10.55), the dual-stack network adaptive scheduling system updates the last probe timestamp field to the current system time "2023-08-21T10:30:45.123Z". This process uses a high-precision clock source (error less than 1 millisecond), and the timestamp format follows the ISO 8601 standard, with additional time zone offset information to adapt to the network environment of multinational enterprises.

[0134] Step 30512: Dynamically adjust the preset re-probing period of the periodic probe timer according to the change ratio of the protocol stack support status flag value in the to-be-re-probed queue.

[0135] In this embodiment, the periodic adjustment algorithm of the dual-stack network adaptive scheduling system optimizes the probe parameters according to the state change frequency. When it is detected that the protocol stack support status of more than 30% of the clients in the to-be-re-probed queue has changed, the dual-stack network adaptive scheduling system shortens the preset re-probing period from 24 hours to 12 hours. The adjustment process of this step adopts an exponential backoff strategy. When a high change rate is detected in three consecutive periods, the period is gradually reduced to the minimum threshold of 6 hours. The dual-stack network adaptive scheduling system also records the adjustment log for the network administrator to audit the effectiveness of the periodic optimization strategy.

[0136] Step 30513: If the change ratio exceeds the dynamic adjustment threshold, shorten the preset re-detection period of the periodic detection timer; if the change ratio is lower than the stability threshold, extend the preset re-detection period of the periodic detection timer.

[0137] In this embodiment, the dynamic adjustment module of the dual-stack network adaptive scheduling system implements optimization of periodic parameters. When the enterprise network is upgraded and 70% of the clients enable the IPv6 protocol stack, the dual-stack network adaptive scheduling system detects that the state change ratio exceeds the dynamic adjustment threshold and immediately adjusts the preset re-detection period from 24 hours to 8 hours. Conversely, when the state change ratio is lower than the stability threshold of 5% for five consecutive cycles, the dual-stack network adaptive scheduling system gradually extends the period to 48 hours. This adjustment process uses a PID control algorithm to smoothly transition the period value to avoid sudden changes in network detection traffic.

[0138] Step 30514: Reset the periodic detection timer and start a new round of periodic re-detection process.

[0139] In this embodiment, the timer control unit of the dual-stack network adaptive scheduling system completes the cycle reset operation. After updating the protocol stack support status of all clients, the dual-stack network adaptive scheduling system sends a timer reset instruction to the core DNS server and each regional proxy node. The new detection period starts counting down based on the current system time, and at the same time, the processed records in the re-detection queue to be processed are cleared. Among them, the reset process uses a two-phase commit protocol to ensure that the timer states of all network nodes are strictly synchronized, establishing an accurate time reference for the next round of periodic re-detection.

[0140] The periodic re-detection mechanism in the above technical solution can build a dynamically adaptive protocol stack status maintenance system through multi-layer collaborative control. In the enterprise network environment, it effectively solves the problem of compatibility status drift caused by device protocol stack upgrades and network configuration changes. By intelligently adjusting the detection frequency and accurately updating the status data, the dual-stack network adaptive scheduling system can ensure that the DNS resolution policy is always based on the latest network topology and terminal capability information, significantly improving the reliability of the dual-stack network service and the resource scheduling efficiency, and providing continuous monitoring guarantee for the smooth evolution of the enterprise network.

[0141] In an alternative implementation, the method further includes:

[0142] Step 401: Configure a global traffic scheduling policy template in the DNS server, and the global traffic scheduling policy template includes protocol stack scheduling rules corresponding to domain names of different service levels.

[0143] In step 401, the policy configuration module of the dual-stack network adaptive scheduling system deploys a global traffic scheduling policy template on the enterprise core DNS server. This template defines the mapping relationship between business levels and protocol stack scheduling through an XML format file, and contains the classification information of all registered domain names in the enterprise office network.

[0144] Taking the enterprise resource planning system domain name "erp.example.com" as an example, it is classified into the first-priority business category, and the corresponding protocol stack scheduling rule forcibly requires the priority use of IPv6 addresses. The template is stored in a highly available distributed database cluster and is visually configured through the graphical interface of the enterprise network management platform, supporting batch import of business domain name lists and hierarchical policy parameters.

[0145] Step 402: In response to the target domain name belonging to the first-priority business category, forcibly set the IPv6 address as the primary address and preferentially allocate it.

[0146] In step 402, the priority execution module of the dual-stack network adaptive scheduling system implements a forced scheduling policy for the first-priority business domain names. When it detects that the video conferencing system domain name "meet.example.com" belongs to the first-priority category, the system ignores the real-time link quality assessment result and directly sets the IPv6 address 2001:db8::1:20 as the primary address. This policy is achieved by modifying the resource record sorting in the DNS response message, placing the AAAA record before the A record. During the forced execution process, the dual-stack network adaptive scheduling system synchronously sends a policy effectiveness notification to the enterprise software-defined network controller, triggering the update of the IPv6 traffic priority routing rule for the corresponding business domain name.

[0147] Step 403: Real-time monitor the network congestion metrics of the IPv6 link. If the network congestion metrics exceed the preset warning threshold, start the protocol stack fallback mechanism.

[0148] In step 403, the network status monitoring service of the dual-stack network adaptive scheduling system real-time collects the congestion metrics of the IPv6 link. For the IPv6 backbone link of the file transfer service domain name "ftp.example.com", the dual-stack network adaptive scheduling system uses the sFlow sampling technology deployed on the core switch to count the packet retransmission rate and buffer queue depth within every 5 seconds. When it detects that the packet retransmission rate breaks through the preset warning threshold of 15%, the dual-stack network adaptive scheduling system generates a protocol stack fallback instruction. The monitoring data is real-time transmitted to the analysis module through the enterprise network operation and maintenance bus and is compared and analyzed with the historical baseline data to accurately identify sudden network congestion events.

[0149] Step 404: Screen available IPv4 addresses from the IP address pool through the protocol stack fallback mechanism, and update the DNS resolution result to replace it with the available IPv4 address.

[0150] In step 404, the address fallback module of the dual-stack network adaptive scheduling system performs a protocol stack switching operation. Taking the enterprise mail system domain name "mail.example.com" as an example, when the IPv6 link is severely congested, the dual-stack network adaptive scheduling system screens out the geographically closest IPv4 address 192.0.2.60 from the pre-configured IP address pool. This address selection process combines the subnet information provided by the client EDNS Client Subnet extension field and matches the optimal access node through the enterprise geographic information database. The updated DNS resolution result is quickly distributed to each regional DNS cache node through anycast addresses to ensure the consistency of the whole network resolution policy.

[0151] Step 405: Write the association relationship between the forced policy and the fallback mechanism into the exception handling rule library of the adaptive scheduling policy.

[0152] In step 405, the rule management unit of the dual-stack network adaptive scheduling system establishes a policy association mechanism. After each protocol stack fallback event occurs, the dual-stack network adaptive scheduling system writes parameters such as the service level, fallback time, and duration of the target domain name "oa.example.com" into the exception handling rule library. This rule library analyzes historical records through machine learning algorithms and automatically adjusts the congestion alarm threshold of the first-priority domain names. For example, when it is detected that a certain domain name frequently triggers a fallback during the business peak period, the system dynamically lowers its packet retransmission rate threshold from 15% to 12% to improve the sensitivity of policy response.

[0153] In a preferred implementation, configuring the global traffic scheduling policy template in DNS server in step 401 includes:

[0154] Step 4011: Create a global traffic scheduling policy template in the policy management module of the DNS server, and the global traffic scheduling policy template contains a service level domain name classification table.

[0155] In step 4011, the template creation service of the dual-stack network adaptive scheduling system initializes the global traffic scheduling policy template in the policy management module. Exemplarily, the enterprise network administrator creates a three-level service classification through the management interface:

[0156] The first priority includes the video conferencing system "meet.example.com" and the core database "db.example.com";

[0157] The second priority covers the file service "ftp.example.com";

[0158] The third priority includes the internal information portal "wiki.example.com".

[0159] Each classification is associated with an independent protocol stack scheduling rule configuration file, which is stored in an encrypted distributed key-value storage system to ensure the security and high availability of policy data.

[0160] Step 4012: Classify different domain names into the first priority, the second priority, and the third priority according to the business level domain name classification table, and bind a preset protocol stack scheduling rule to each business level.

[0161] In step 4012, the business classification module of the dual-stack network adaptive scheduling system implements domain name classification mapping. The domain name of the enterprise office automation system "oa.example.com" is classified into the first priority category, and its corresponding protocol stack scheduling rule is configured to force IPv6 priority and enable real-time monitoring. The domain name of the customer relationship management system "crm.example.com" in the second priority sets a dynamic quality optimization strategy, and the domain name of the employee forum "bbs.example.com" in the third priority adopts a load balancing mode. The classification process matches the domain name characteristics through regular expressions and supports wildcard rules to achieve batch classification.

[0162] Step 4013: For the domain names of the first priority business category, set the forced use of the IPv6 address as the primary address in the protocol stack scheduling rule, and activate the enable switch of the protocol stack fallback mechanism.

[0163] In step 4013, the rule binding module of the dual-stack network adaptive scheduling system configures a forced policy for high-priority domain names. For the domain name of the enterprise unified communication system "uc.example.com", the dual-stack network adaptive scheduling system sets the primary address type to IPv6 in the protocol stack scheduling rule and activates the monitoring switch of the protocol stack fallback mechanism. This configuration is achieved by modifying the binary flag bits of the policy template. During the effective period of the forced policy, even if the IPv6 link quality score is lower than that of IPv4, the dual-stack network adaptive scheduling system still preferentially returns AAAA records. The enable switch of the fallback mechanism is linked with the network probe service to monitor the IPv6 service availability of the target domain name in real time.

[0164] Step 4014: Real-time collect the real-time congestion parameters of the IPv6 link corresponding to the target domain name, and the real-time congestion parameters include the packet retransmission rate and the buffer queue depth.

[0165] In step 4014, the performance collection unit of the dual-stack network adaptive scheduling system obtains IPv6 link status parameters in real time. Taking the enterprise cloud storage service domain name "cloud.example.com" as an example, the dual-stack network adaptive scheduling system counts the number of IPv6 packet retransmissions and the buffer queue occupancy rate every 30 seconds through the NetFlow collector deployed on the border router. The collected data is aggregated and then stored in the time series database to provide a real-time data source for congestion detection. The buffer queue depth in this step is measured by combining active probing and passive monitoring, and the end-to-end delay fluctuation is calculated by sending probing packets with timestamps.

[0166] Step 4015: When the real-time congestion parameter exceeds the preset warning threshold in the protocol stack scheduling rule, trigger the startup condition of the protocol stack fallback mechanism.

[0167] In step 4015, the threshold detection module of the dual-stack network adaptive scheduling system triggers the protocol stack fallback mechanism. When it is monitored that the IPv6 link packet retransmission rate of the video conferencing domain name "meet.example.com" reaches 18% (exceeding the preset threshold of 15%), the dual-stack network adaptive scheduling system generates a level-three warning event. The event processing module immediately interrupts the current scheduling policy and sends a protocol stack switching request to the address selection service. The switching decision process introduces a progressive response mechanism. At the initial stage of threshold breakthrough, only some clients are reverted, and the reverted scope is gradually expanded as the congestion level intensifies.

[0168] Step 4016: According to the protocol stack fallback mechanism, screen out the set of IPv4 addresses bound to the target domain name from the IP address pool.

[0169] In step 4016, the address screening service of the dual-stack network adaptive scheduling system obtains standby IPv4 addresses from the IP address pool. For the enterprise VPN access domain name "vpn.example.com", the dual-stack network adaptive scheduling system queries the pre-configured address pool to obtain three available IPv4 addresses: 192.0.2.101, 192.0.2.102, 192.0.2.103. The screening process in this step performs strict health checks, including ICMP reachability test, TCP port response verification, and application layer service detection. The addresses passing multi-dimensional verification are marked as candidate fallback targets and stored in the temporary address cache queue.

[0170] Step 4017: Based on the geographical proximity index of each IPv4 address in the set of IPv4 addresses, select the IPv4 address closest to the geographical location of the client as the fallback target address.

[0171] In step 4017, the geographical matching algorithm of the dual-stack network adaptive scheduling system selects the optimal fallback address. When the client of a city branch accesses "file.example.com", the dual-stack network adaptive scheduling system selects 192.0.2.105 located in the data center of the eastern region from the candidate IPv4 address pool according to the location information of the client IP address. The geographical proximity index is calculated by measuring the number of autonomous system (AS) path hops between the client and each data center, and the access node of the same city and the same operator is preferentially selected. The matching process of this step refers to the optimal path mark in the enterprise network topology diagram to ensure the optimal network performance after fallback.

[0172] Step 4018: Encapsulate the fallback target address into the preferred record field of the DNS response message, and generate a protocol stack fallback operation log containing the handover timestamp and congestion parameters.

[0173] In step 4018, the response reconstruction module of the dual-stack network adaptive scheduling system generates a new DNS resolution result. For the fallback target address 192.0.2.108, the dual-stack network adaptive scheduling system encapsulates it at the beginning of the AnswerSection of the DNS response message, and adds the protocol stack fallback identifier 0xA5 in the extension field. The response message is digitally signed by the enterprise intranet DNS cluster and then sent to the client. The dual-stack network adaptive scheduling system synchronously generates a detailed operation log, recording parameters such as the fallback time, duration, and the number of affected clients. The log entries are transmitted to the centralized audit platform through a secure channel.

[0174] Step 4019: Associate the protocol stack fallback operation log with the service level of the target domain name, and update it to the exception handling rule library of the global traffic scheduling policy template; dynamically adjust the preset alarm threshold in the protocol stack scheduling rule according to the historical fallback records of domain names with the same service level in the exception handling rule library.

[0175] In step 4019, the policy optimization module of the dual-stack network adaptive scheduling system realizes closed-loop control. After each protocol stack fallback event, the dual-stack network adaptive scheduling system analyzes the historical records of "erp.example.com" and finds that there have been 8 fallbacks caused by the excessive packet retransmission rate in the past 30 days. Based on this, the dual-stack network adaptive scheduling system automatically adjusts the retransmission rate threshold of this domain name from 15% to 12%. The above optimization process uses a sliding window algorithm, focuses on the operation data of the most recent 7 days, and introduces holiday traffic pattern recognition to dynamically adapt to the periodic changes of enterprise services.

[0176] In an optional implementation, the protocol stack fallback mechanism in step 403 includes step 4030: collecting real-time congestion parameters of the IPv6 link, where the real-time congestion parameters include packet retransmission rate, buffer overflow count, and average queuing delay; if the packet retransmission rate exceeds a first congestion threshold, or the buffer overflow count exceeds a second congestion threshold, then determine that the IPv6 link is in an unavailable state; select a set of IPv4 addresses bound to the target domain name from the IP address pool, and filter the target IPv4 address based on geographical proximity; generate a protocol stack fallback instruction, and encapsulate the target IPv4 address into the preferred record field of the DNS response message; send a protocol stack fallback warning message to the network management end, and record the log data of the fallback operation.

[0177] In step 4030, the protocol stack fallback mechanism of the dual-stack network adaptive scheduling system performs a full process operation. When it is monitored that the buffer overflow count of the IPv6 link of the enterprise unified communication system domain name "uc.example.com" accumulates to 50 times within 5 minutes (exceeding the second congestion threshold of 40 times), the dual-stack network adaptive scheduling system determines that this link is in an unavailable state. The address selection service immediately selects two IPv4 addresses 192.0.2.110 and 192.0.2.115 in the eastern and northern regions from the IP address pool, and selects the optimal address based on the geographical location of the client. After encapsulating the fallback instruction into the DNS response message, the dual-stack network adaptive scheduling system sends an SNMP trap warning message to the network operation and maintenance center, and details and records the event time, affected scope, and solution. The entire above-mentioned fallback process can be completed within a few hundred milliseconds, minimizing the service interruption time.

[0178] As an optional but non-limiting embodiment, generating an adaptive scheduling policy including a dynamic traffic splitting ratio of IPv4 addresses and IPv6 addresses based on the adjusted sorting weights includes:

[0179] Step 10351: Obtain the final sorting weights of the IPv4 link quality score and the IPv6 link quality score, calculate the weight proportion of the IPv4 address and the weight proportion of the IPv6 address; according to the ratio of the weight proportion of the IPv4 address to the weight proportion of the IPv6 address, determine the dynamic traffic splitting ratio base of the IPv4 address and the IPv6 address.

[0180] In step 10351, the weight calculation module of the dual-stack network adaptive scheduling system generates a dynamic traffic splitting ratio base according to the final sorting weights. Taking the enterprise video conferencing system domain name "meet.example.com" as an example, the dual-stack network adaptive scheduling system obtains the final sorting weights of the IPv4 link quality score of 85 points and the IPv6 link quality score of 92 points, where the IPv4 weight accounts for 48% and the IPv6 weight accounts for 52%. By calculating the ratio of the two, 0.923 (48 / 52), the initial dynamic traffic splitting ratio base is determined as IPv4:IPv6 = 48:52. This calculation process uses a normalization algorithm to map the scores of each protocol stack to the weight range of 0-100% to ensure the comparability of the ratio base. The dual-stack network adaptive scheduling system synchronously verifies the rationality of the weight ratio and triggers an artificial review process when abnormal fluctuations in the weight of a certain protocol stack are detected.

[0181] Step 10352: Extract the difference in the successful connection rates of IPv4 addresses and IPv6 addresses in the historical address allocation records corresponding to the target domain name, and calculate the offset of the difference in the successful connection rates from the preset reference difference; generate a correction coefficient for the dynamic traffic splitting ratio base according to the positive or negative sign and the absolute value of the offset; multiply the correction coefficient by the dynamic traffic splitting ratio base to obtain the corrected dynamic traffic splitting ratio of IPv4 addresses and IPv6 addresses; based on the corrected dynamic traffic splitting ratio, determine the primary allocation interval and the backup allocation interval of IPv4 addresses and IPv6 addresses according to the ratio threshold interval division rule.

[0182] In step 10352, the historical data analysis module of the dual-stack network adaptive scheduling system extracts the difference in the connection success rates of the target domain name for ratio correction. For the enterprise file service domain name "file.example.com", the dual-stack network adaptive scheduling system retrieves the difference of 1.3% between the IPv4 address connection success rate of 98.2% and the IPv6 address connection success rate of 99.5% in the past 24 hours, and generates a positive offset of 0.8% compared with the preset reference difference of 0.5%. A correction coefficient of +1.5% is generated according to the absolute value of the offset, and the initial ratio base IPv4:IPv6 = 45:55 is adjusted to the corrected 43.5:56.5. The correction process uses a piecewise linear interpolation algorithm, and an exponential compensation mechanism is enabled when the offset exceeds 2%. The dual-stack network adaptive scheduling system divides the primary allocation interval (≥55%) and the backup allocation interval (<45%) according to the corrected ratio, and reserves an elastic adjustment space in the middle interval.

[0183] Step 10353: When the proportion of IPv4 addresses in the dynamic traffic splitting ratio is within the primary allocation range, set the IPv4 address as the first-priority allocation address; when the proportion of IPv6 addresses in the dynamic traffic splitting ratio is within the primary allocation range, set the IPv6 address as the first-priority allocation address.

[0184] In step 10353, the priority decision module of the dual-stack network adaptive scheduling system determines the primary address type. When the corrected result of the dynamic traffic splitting ratio of the enterprise unified communication system domain name "uc.example.com" shows that the IPv6 proportion is 58%, the dual-stack network adaptive scheduling system sets the IPv6 address 2001:db8::1:30 as the primary allocation address type. This setting is achieved by modifying the protocol stack priority field in the DNS policy library, and the AAAA record is preferentially returned during subsequent address allocation. For the traffic splitting ratio in the intermediate range (such as IPv4:IPv6 = 50:50), the dual-stack network adaptive scheduling system defaults to maintaining the current primary protocol stack type until the next evaluation cycle triggers a recalculation.

[0185] Step 10354: According to the address type of the primary allocation range, screen out the address set of the corresponding protocol stack type from the IP address pool; according to the ratio difference between the primary allocation range and the secondary allocation range, adjust the address sorting of different geographical location nodes in the address set.

[0186] In step 10354, the address screening module of the dual-stack network adaptive scheduling system performs geographical location optimization sorting. Taking the access of enterprise branch offices to "oa.example.com" as an example, when the primary allocation range is determined to be IPv6, the dual-stack network adaptive scheduling system selects three IPv6 addresses from the IP address pool: 2001:db8::1:10 (eastern node), 2001:db8::1:11 (northern node), and 2001:db8::1:12 (southern node). After identifying the client's location as the eastern region based on the EDNS subnet information of the client, the dual-stack network adaptive scheduling system raises the sorting of 2001:db8::1:10 to the top. The sorting adjustment process refers to the autonomous system path cost data in the enterprise network topology diagram, and preferentially selects service nodes with a direct backbone link to the client.

[0187] Step 10355: Package the sorted address set into different record fields of the DNS response message in the order of priority; select the address closest to the geographical location of the client from the sorted address set as the preferred resolution result; add the preferred resolution result and the secondary resolution result to the address allocation queue of the DNS response message; set the quantity distribution of addresses with different protocol stacks in the address allocation queue according to the ratio of the primary address to the standby address in the dynamic traffic splitting ratio.

[0188] In step 10355, the response encapsulation module of the dual-stack network adaptive scheduling system constructs an address allocation queue. For the dynamic traffic splitting ratio of the enterprise cloud storage domain name "cloud.example.com" with IPv4:IPv6 = 40:60, the dual-stack network adaptive scheduling system fills in the IPv6 addresses 2001:db8::1:20 and 2001:db8::1:21 in the first two positions of the DNS response message Answer Section, and fills in the IPv4 addresses 192.0.2.30, 192.0.2.31, and 192.0.2.32 in the next three positions. According to the geographical location information of the eastern client, the dual-stack network adaptive scheduling system sets the IPv6 address 2001:db8::1:20 located in a data center in a certain city as the preferred resolution result. The quantity distribution of the address queue strictly follows the ratio requirement of 6:4, and the IPv6 addresses occupy the first six sorting positions in 10 resource records.

[0189] Step 10356: In response to the client initiating a connection request, return the resolution results of addresses with different protocol stacks in sequence according to the priority order of the address allocation queue.

[0190] In step 10356, the request processing unit of the dual-stack network adaptive scheduling system returns the resolution results in the order of priority. When the enterprise mobile office client accesses "crm.example.com", the dual-stack network adaptive scheduling system sequentially returns the IPv6 addresses 2001:db8::1:40, 2001:db8::1:41 and the IPv4 address 192.0.2.50 according to the address allocation queue. The client application layer protocol stack attempts to establish a connection according to the DNS response order, and first initiates a TCP three-way handshake for the first IPv6 address. The dual-stack network adaptive scheduling system records the actual address type used through the connection tracking module deployed on the core switch. When it detects that a certain address fails to connect three times in a row, it automatically degrades its sorting in the queue and triggers an address health check process.

[0191] Based on the same inventive concept, the embodiment of the present application also provides a dual-stack network adaptive scheduling system. Refer to Figure 2As shown, it is a schematic structural diagram of a possible dual-stack network adaptive scheduling system provided in an embodiment of the present application. Figure 2 In the dual-stack network adaptive scheduling system 200, it includes: a processor 210 and a memory 220. Among them, the memory 220 stores a computer program executable by the processor 210. By executing the instructions stored in the memory 220, the processor 210 can execute the steps of the above-mentioned dual-stack network adaptive scheduling method based on the DNS resolution policy.

[0192] Based on the same inventive concept, an embodiment of the present application provides a computer-readable storage medium, which includes a computer program. When the computer program runs on the dual-stack network adaptive scheduling system, the computer program is used to make the dual-stack network adaptive scheduling system execute the steps of the above-mentioned dual-stack network adaptive scheduling method based on the DNS resolution policy. In some possible implementation manners, various aspects of the dual-stack network adaptive scheduling method provided in the present application can also be implemented in the form of a program product, which includes a computer program. When the program product runs on the dual-stack network adaptive scheduling system, the computer program is used to make the dual-stack network adaptive scheduling system execute the steps in the above-mentioned dual-stack network adaptive scheduling method based on the DNS resolution policy. For example, the dual-stack network adaptive scheduling system can execute steps as Figure 1 shown in.

Claims

1. A dual-stack network adaptive scheduling method based on DNS resolution strategy, characterized in that Including: Receiving a DNS resolution request sent by a client, where the DNS resolution request carries a target domain name and a request protocol type; Parsing the request protocol type to determine the network protocol stack type supported by the client; Generating an adaptive scheduling policy corresponding to the target domain name according to the network protocol stack type and a preset address type priority list: obtaining a first set of real-time parameters of the IPv4 network link associated with the target domain name in real time, where the first set of real-time parameters includes IPv4 link latency, IPv4 node available bandwidth, and IPv4 request response success rate; Synchronously collecting a second set of real-time parameters of the IPv6 network link associated with the target domain name, where the second set of real-time parameters includes IPv6 link throughput, IPv6 node load balancing metrics, and IPv6 path hop count; Comparing the first set of real-time parameters with preset IPv4 baseline parameters to generate an IPv4 link quality score; Comparing the second set of real-time parameters with preset IPv6 baseline parameters to generate an IPv6 link quality score; dynamically adjusting the sorting weights of IPv4 and IPv6 in the address type priority list according to the difference between the IPv4 link quality score and the IPv6 link quality score; Based on the adjusted sorting weights, generating an adaptive scheduling policy including a dynamic splitting ratio of IPv4 addresses and IPv6 addresses, where the dynamic splitting ratio is used to control the address allocation priority of different protocol stacks, and the dynamic splitting ratio is used to indicate the address allocation ratio of a mixed resolution result including IPv4 addresses and IPv6 addresses; Selecting a target IP address type from a pre-configured IP address pool based on the adaptive scheduling policy; Returning the resolution result corresponding to the target IP address type to the client to guide the client to establish a network connection through the target IP address type.

2. The method according to claim 1, characterized in that The dynamically adjusting the sorting weights of IPv4 and IPv6 in the address type priority list according to the difference between the IPv4 link quality score and the IPv6 link quality score includes: If the IPv4 link quality score is higher than the IPv6 link quality score and the difference exceeds a first threshold, then increasing the sorting level of IPv4 in the address type priority list; if the IPv6 link quality score is higher than the IPv4 link quality score and the difference exceeds a second threshold, then increasing the sorting level of IPv6; otherwise, keeping the current sorting weights unchanged; Compensating and correcting the sorting weights according to the successful connection rates of IPv4 and IPv6 in the historical scheduling records; Weightedly fusing the compensated and corrected sorting weights with the real-time link quality score to generate a final sorting weight to update the address type priority list; The compensating and correcting the sorting weights according to the successful connection rates of IPv4 and IPv6 in the historical scheduling records includes: Extracting the ratio of the cumulative number of successful connections to the number of failed connections of IPv4 addresses under the target domain name within a preset historical time period as the IPv4 historical success rate; Extract the ratio of the cumulative number of successful connections to the number of failed connections of IPv6 addresses within the same time period as the IPv6 historical success rate; Calculate the ratio difference between the IPv4 historical success rate and the IPv6 historical success rate; If the ratio difference exceeds the preset compensation threshold, increase the difference compensation coefficient for the current sorting weight; Linearly superimpose the difference compensation coefficient and the real-time link quality score to generate a compensated sorting weight.

3. The method according to claim 1, wherein The selecting of the target IP address type from the pre-configured IP address pool based on the adaptive scheduling strategy includes: When the adaptive scheduling strategy indicates to adopt the protocol stack parallel mode, simultaneously extract the IPv4 address list and the IPv6 address list from the IP address pool; Determine the primary allocation order and the secondary allocation order of the IPv4 and IPv6 addresses according to the request protocol type of the client; Encapsulate the IPv4 or IPv6 address corresponding to the primary allocation order into the preferred record field of the DNS response message; Add the secondary protocol stack address corresponding to the secondary allocation order to the additional record field of the DNS response message; In response to the failure of the client to connect to the primary address, trigger a protocol stack switching process based on the secondary address in the additional record field to guide the client to re-initiate a connection request; The protocol stack switching process includes: Monitor the connection establishment status between the client and the primary address. If no connection confirmation signal is received within the preset timeout period, generate a protocol stack switching instruction; Parse the additional record field of the DNS response message to extract the protocol stack type and IP address corresponding to the secondary address; Send an address redirection message to the client, where the address redirection message includes the secondary address and a protocol stack switching identifier; Record the protocol stack type, switching timestamp, and switching reason of this switching operation and update them to the historical switching log; Dynamically adjust the allocation ratio of the primary and secondary addresses in the subsequent scheduling strategy according to the cumulative number of switches of the same target domain name in the historical switching log.

4. The method according to claim 1, wherein The method further includes: Deploy a protocol stack compatibility detection service at the DNS resolution layer. The following operations are performed by the protocol stack compatibility detection service: Send an IPv4 protocol detection message and an IPv6 protocol detection message to the client. Both the IPv4 protocol detection message and the IPv6 protocol detection message include a preset protocol identifier; Listen to the feedback behavior of the client on the IPv4 protocol detection message and the IPv6 protocol detection message within the preset response time; If only the IPv4 detection response is received, mark the protocol stack support status of the client as only IPv4 compatible; If only the IPv6 detection response is received, mark the protocol stack support status of the client as only IPv6 compatible; If both the IPv4 detection response and the IPv6 detection response are received, mark the protocol stack support status of the client as dual-stack compatible; Store the marking result in the protocol stack support status table, and the protocol stack support status table is the direct call object for subsequent DNS resolution requests; The calling process of the protocol stack support status table includes: When receiving a new DNS resolution request, extract the source IP address and request protocol type of the new client corresponding to the new DNS resolution request; Query the historical marking result associated with the source IP address in the protocol stack support status table; If the historical marking result indicates that the new client is dual-stack compatible, allocate an IPv4 address or an IPv6 address according to the current adaptive scheduling policy; If the historical marking result is single-protocol-stack compatible, skip the adaptive scheduling policy generation step and directly allocate the corresponding protocol stack address; When the protocol stack support status of the new client changes, update the protocol stack support status table through periodic re-probing.

5. The method according to claim 4, wherein The step of updating the protocol stack support status table through periodic re-probing when the protocol stack support status of the new client changes includes: Deploy a timer trigger mechanism at the DNS resolution layer and start a periodic detection timer associated with the protocol stack support status table; When the periodic detection timer reaches the preset re-probing period, scan the protocol stack support status entries of all clients in the protocol stack support status table; Extract the last detection timestamp and protocol stack support status marking value corresponding to the source IP address of each client; Calculate the time interval between the current system time and the last detection timestamp, and compare the difference with the preset re-probing period; If the time interval exceeds the validity threshold of the preset re-probing period, add the source IP address of the client to the re-probing queue; According to the order of the source IP addresses in the re-probing queue, send an IPv4 protocol detection message and an IPv6 protocol detection message to each client; Listen for the type of detection response message returned by the client within the preset response time window; Extract the protocol identifier and client source IP address carried in the detection response message; If the protocol identifier indicates that the client returns an IPv4 detection response, update the IPv4 compatibility mark of the corresponding source IP address in the protocol stack support status table to the valid state; if the protocol identifier indicates that the client returns an IPv6 detection response, update the IPv6 compatibility mark of the corresponding source IP address in the protocol stack support status table to the valid state; if no IPv4 detection response or IPv6 detection response is received within the response time window, set the corresponding protocol stack compatibility mark to the invalid state; According to the updated protocol stack compatibility mark, recalculate the protocol stack support status marking value of the client; overwrite the historical marking value in the protocol stack support status table with the recalculated protocol stack support status marking value; Record the timestamp of this re-probing operation to the last detection timestamp field of the protocol stack support status table; Dynamically adjust the preset re-probing period of the periodic detection timer according to the change ratio of the protocol stack support status marking value in the re-probing queue; If the change ratio exceeds the dynamic adjustment threshold, shorten the preset re-probing period of the periodic detection timer; if the change ratio is lower than the stable threshold, extend the preset re-probing period of the periodic detection timer; Reset the periodic detection timer and start a new round of periodic re-detection process.

6. The method according to claim 1, wherein The method further includes: Configure a global traffic scheduling policy template in the DNS server, where the global traffic scheduling policy template includes protocol stack scheduling rules corresponding to domain names of different service levels; In response to the target domain name belonging to the first priority service category, force the IPv6 address to be set as the primary address and allocate it preferentially; Real-time monitor the network congestion metrics of the IPv6 link. If the network congestion metrics exceed the preset warning threshold, start the protocol stack fallback mechanism; Screen available IPv4 addresses from the IP address pool through the protocol stack fallback mechanism, and update the DNS resolution result to replace it with the available IPv4 address; Write the association relationship between the forced policy and the fallback mechanism into the exception handling rule library of the adaptive scheduling policy.

7. The method according to claim 6, characterized in that, The configuring of the global traffic scheduling policy template in the DNS server includes: Create a global traffic scheduling policy template in the policy management module of the DNS server, where the global traffic scheduling policy template contains a service level domain name classification table; According to the service level domain name classification table, divide different domain names into the first priority, the second priority, and the third priority according to the service level, and bind a preset protocol stack scheduling rule to each service level; For the domain name of the first priority service category, set the forced use of the IPv6 address as the primary address in the protocol stack scheduling rule, and activate the enabling switch of the protocol stack fallback mechanism; Real-time collect the real-time congestion parameters of the IPv6 link corresponding to the target domain name, where the real-time congestion parameters include the packet retransmission rate and the buffer queue depth; When the real-time congestion parameters exceed the preset warning threshold in the protocol stack scheduling rule, trigger the start condition of the protocol stack fallback mechanism; According to the protocol stack fallback mechanism, screen the set of IPv4 addresses bound to the target domain name from the IP address pool; Based on the geographical proximity metrics of each IPv4 address in the set of IPv4 addresses, select the IPv4 address closest to the client's geographical location as the fallback target address; Encapsulate the fallback target address into the preferred record field of the DNS response message, and generate a protocol stack fallback operation log containing the handover timestamp and congestion parameters; Associate the protocol stack fallback operation log with the service level of the target domain name, and update it to the exception handling rule library of the global traffic scheduling policy template; Dynamically adjust the preset warning threshold in the protocol stack scheduling rule according to the historical fallback records of domain names of the same service level in the exception handling rule library.

8. The method according to claim 6, wherein The protocol stack fallback mechanism includes: Collect the real-time congestion parameters of the IPv6 link, where the real-time congestion parameters include the packet retransmission rate, the number of buffer overflows, and the average queuing delay; If the packet retransmission rate exceeds the first congestion threshold, or the number of buffer overflows exceeds the second congestion threshold, determine that the IPv6 link is in an unavailable state; Select a set of IPv4 addresses bound to the target domain name from the IP address pool, and screen the target IPv4 address based on geographical proximity; Generate a protocol stack rollback instruction and encapsulate the target IPv4 address into the preferred record field of the DNS response message; Send a protocol stack rollback warning message to the network management terminal and record the log data of the rollback operation.

9. A dual-stack network adaptive scheduling system, characterized in that, It includes a processor and a memory. Among them, the memory stores a computer program. When the computer program is executed by the processor, the processor is caused to execute the steps of any one of the methods recited in claims 1 to 8.

Images