A resource scheduling method and device based on service measurement, medium and product

Abstract

The application discloses a resource scheduling method and device based on service measurement, a medium and a product. The method comprises the following steps: collecting scheduling resource data of regional traffic service scheduling; analyzing service quality and traffic flow direction of the scheduling resource data to determine evaluation indexes; predicting bandwidth utilization and service quality indexes of the next period according to the evaluation indexes and scoring respectively; determining the distribution weight of local IDC analysis fast table according to the scoring result; selecting the local IDC analysis fast table for scheduling query according to the distribution weight to determine the resource scheduling address. The application can improve the resource scheduling accuracy and efficiency and improve the service quality of user services.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and more specifically, to a resource scheduling method, apparatus, medium, and product based on service calculation. Background Technology

[0002] CDN (Content Delivery Network) is a widely used technology in network services that improves the speed and quality of website access for users by distributing content to nodes around the world. In this network, GSLB (Global Server Load Balancing) is typically implemented through Intelligent DNS (Domain Name Server). Based on the address of the Local DNS (Local Domain Name Server) configured by the customer, content is distributed to the nearest available cache server. Therefore, the accuracy of CDN scheduling relies on the intelligent resolution of GSLB.

[0003] Existing technologies acquire CDN operational status data and schedule resources based on server availability and service quality, such as CPU, memory, and hard drive capacity. However, existing scheduling strategies simply calculate server availability without dynamically weighting resource scheduling based on end-to-end service quality expectations and anticipated resource load. This results in low accuracy and efficiency in resource scheduling, leading to lower service quality for users. Summary of the Invention

[0004] Compared with existing technologies, this invention proposes a resource scheduling method, device, medium, and product based on business calculation, which can improve the accuracy and efficiency of resource scheduling and improve the quality of user business services.

[0005] This invention provides a resource scheduling method based on business calculations, the method comprising:

[0006] Collect scheduling resource data for traffic service scheduling in this region;

[0007] The scheduling resource data is analyzed for service quality and traffic flow to determine evaluation indicators;

[0008] Based on the aforementioned evaluation indicators, predict the bandwidth utilization and service quality indicators for the next cycle, and score them accordingly.

[0009] The allocation weight of the local IDC parsing fast table is determined based on the scoring results;

[0010] Based on the assigned weights, the local IDC parsing fast table is selected for scheduling queries to determine the resource scheduling address.

[0011] Preferably, the data collected includes scheduling resource data for traffic service scheduling in this region, including:

[0012] Obtain GSLB scheduling data, DNS-related IDC IP resolution data, regional network port outbound DPI traffic data, and IDC outbound DPI traffic data for traffic services in this region;

[0013] Perform synchronous backup of the GSLB scheduling data;

[0014] The acquired data is format-converted and grouped to synthesize an XDR call detail record.

[0015] Furthermore, the scheduling resource data is analyzed for service quality and traffic flow to determine evaluation indicators, including:

[0016] The drill-down levels for different IDC customers, different services, and different domain names are determined based on the customer information, IP address database information, and domain name information in the scheduling resource data.

[0017] By conducting specific business quality and domain name quality analyses at different drilling levels, business quality indicators are determined.

[0018] By analyzing the service traffic flow rate of websites at different drill-down levels, the bandwidth utilization of IDC-related links can be determined.

[0019] The service quality indicators and the bandwidth utilization rate are used as the evaluation indicators.

[0020] Furthermore, define the business quality indicators, including:

[0021] The system associates the pre-defined IP address database with the IPs drilled down by the SPs in the business category, analyzes and identifies high-quality and low-quality IPs, marks their respective subdomains, and determines the IPs' HTTP response success rate, HTTP response latency, TCP establishment success rate, and TCP establishment latency as business metrics.

[0022] The preset domain name database, the IP address database, and the second-level domain names drilled down by the SP of the business category are correlated to analyze high-quality domain names and low-quality domain names. The SP and IP of the second-level domain name are labeled, and the HTTP response success rate, HTTP response latency, TCP establishment success rate, and TCP establishment latency of the second-level domain name are determined as business indicators.

[0023] As a preferred approach, determine the bandwidth utilization of IDC-related links, including:

[0024] The system associates the preset IP address database with the first-level domains drilled down by the SP for major business categories to generate peak flow rates and load conditions for each first-level domain of important websites to each region, as well as other downlink peak flow rates for IDC SP services to other regions and other customers.

[0025] The IP address database and the subdomains drilled down from the first-level domain are associated to generate peak flow rates and load conditions for each subdomain under each first-level domain of important websites in each region, as well as other downlink peak flow rates for IDC SP services in other regions and for other customers;

[0026] The bandwidth utilization of IDC-related links is determined based on traffic analysis.

[0027] Preferably, the bandwidth utilization rate and service quality indicators for the next cycle are predicted based on the evaluation indicators, and scores are assigned to each, including:

[0028] The bandwidth utilization rate and service quality indicators collected during this period are calculated to determine the stability value, trend value, and seasonality value.

[0029] The three-stage exponential smoothing method is used to calculate the stationarity, trend, and seasonality values ​​of bandwidth utilization and service quality indicators, respectively, to predict the bandwidth utilization of IP and related service indicators of IP in the next cycle.

[0030] The bandwidth utilization rate and related business indicators of the IP in the next cycle are scored according to the pre-defined scoring intervals.

[0031] Preferably, the allocation weight of the local IDC parsing fast table is determined based on the scoring results, including:

[0032] The weights of each IP address are determined by weighted summation of the scores for bandwidth utilization and service quality indicators for the same IP address.

[0033] Sum the IP weights of all IPs under the same domain name and update the assigned weights of different domains in the local IDC resolution table.

[0034] Preferably, the local IDC parsing TLB is selected for scheduling query based on the allocation weight to determine the resource scheduling address, including:

[0035] Perform the lookup based on the local IDC parsing fast table;

[0036] When a resource address is found and the corresponding allocation weight is not 0, it is fed back to LDNS for resource scheduling.

[0037] When no resource address is found, or the allocation weight corresponding to the found resource address is 0, the GSLB scheduling data in the scheduling resource data is parsed and searched, and the latest resource address found is fed back to LDNS for resource scheduling.

[0038] This invention also provides a resource scheduling device based on business calculations, the device comprising:

[0039] The data acquisition module is used to collect scheduling resource data for traffic service scheduling in this region;

[0040] The analysis module is used to perform service quality and traffic flow analysis on the scheduling resource data and determine evaluation indicators;

[0041] The prediction module is used to predict the bandwidth utilization rate and service quality indicators for the next period based on the evaluation indicators, and to score them respectively.

[0042] The weighting module is used to determine the allocation weight of the local IDC parsing fast table based on the scoring results;

[0043] The scheduling module is used to select the local IDC parsing fast table for scheduling based on the allocation weight, and determine the resource scheduling address.

[0044] Preferably, the data acquisition module is specifically used for:

[0045] Obtain GSLB scheduling data, DNS-related IDC IP resolution data, regional network port outbound DPI traffic data, and IDC outbound DPI traffic data for traffic services in this region;

[0046] Perform synchronous backup of the GSLB scheduling data;

[0047] The acquired data is format-converted and grouped to synthesize an XDR call detail record.

[0048] Preferably, the analysis module is specifically used for:

[0049] The drill-down levels for different IDC customers, different services, and different domain names are determined based on the customer information, IP address database information, and domain name information in the scheduling resource data.

[0050] By conducting specific business quality and domain name quality analyses at different drilling levels, business quality indicators are determined.

[0051] By analyzing the service traffic flow rate of websites at different drill-down levels, the bandwidth utilization of IDC-related links can be determined.

[0052] The service quality indicators and the bandwidth utilization rate are used as the evaluation indicators.

[0053] Furthermore, the analysis module is specifically used for:

[0054] The system associates the pre-defined IP address database with the IPs drilled down by the SPs in the business category, analyzes and identifies high-quality and low-quality IPs, marks their respective subdomains, and determines the IPs' HTTP response success rate, HTTP response latency, TCP establishment success rate, and TCP establishment latency as business metrics.

[0055] The preset domain name database, the IP address database, and the second-level domain names drilled down by the SP of the business category are correlated to analyze high-quality domain names and low-quality domain names. The SP and IP of the second-level domain name are labeled, and the HTTP response success rate, HTTP response latency, TCP establishment success rate, and TCP establishment latency of the second-level domain name are determined as business indicators.

[0056] Furthermore, the analysis module is specifically used for:

[0057] The system associates the preset IP address database with the first-level domains drilled down by the SP for major business categories to generate peak flow rates and load conditions for each first-level domain of important websites to each region, as well as other downlink peak flow rates for IDC SP services to other regions and other customers.

[0058] The IP address database and the subdomains drilled down from the first-level domain are associated to generate peak flow rates and load conditions for each subdomain under each first-level domain of important websites in each region, as well as other downlink peak flow rates for IDC SP services in other regions and for other customers;

[0059] The bandwidth utilization of IDC-related links is determined based on traffic analysis.

[0060] Preferably, the prediction module is specifically used for:

[0061] The bandwidth utilization rate and service quality indicators collected during this period are calculated to determine the stability value, trend value, and seasonality value.

[0062] The three-stage exponential smoothing method is used to calculate the stationarity, trend, and seasonality values ​​of bandwidth utilization and service quality indicators, respectively, to predict the bandwidth utilization of IP and related service indicators of IP in the next cycle.

[0063] The bandwidth utilization rate and related business indicators of the IP in the next cycle are scored according to the pre-defined scoring intervals.

[0064] Preferably, the weighting module is specifically used for:

[0065] The weights of each IP address are determined by weighted summation of the scores for bandwidth utilization and service quality indicators for the same IP address.

[0066] Sum the IP weights of all IPs under the same domain name and update the assigned weights of different domains in the local IDC resolution table.

[0067] Preferably, the scheduling module is specifically used for:

[0068] Perform the lookup based on the local IDC parsing fast table;

[0069] When a resource address is found and the corresponding allocation weight is not 0, it is fed back to LDNS for resource scheduling.

[0070] When no resource address is found, or the allocation weight corresponding to the found resource address is 0, the GSLB scheduling data in the scheduling resource data is parsed and searched, and the latest resource address found is fed back to LDNS for resource scheduling.

[0071] This invention also provides a resource scheduling device based on business calculation, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a resource scheduling method based on business calculation as described in any of the above embodiments.

[0072] This invention also provides a computer-readable storage medium, which includes a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform a resource scheduling method based on business calculation as described in any of the above embodiments.

[0073] This invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method described in any of the above embodiments.

[0074] Compared with existing technologies, this invention provides a resource scheduling method, apparatus, medium, and product based on service calculation. It collects scheduling resource data for local traffic service scheduling; analyzes the scheduling resource data for service quality and traffic flow to determine evaluation indicators; predicts bandwidth utilization and service quality indicators for the next cycle based on the evaluation indicators, and scores them respectively; determines the allocation weight of the local IDC parsing fast table based on the scoring results; and selects the local IDC parsing fast table for scheduling queries based on the allocation weight to determine the resource scheduling address. This application's solution can improve resource scheduling accuracy and efficiency, and enhance the quality of user service. Attached Figure Description

[0075] Figure 1 A flowchart illustrating a resource scheduling method based on business calculations provided in this embodiment of the invention;

[0076] Figure 2 This is another flowchart illustrating the resource scheduling method based on business calculation provided in this embodiment of the invention;

[0077] Figure 3 This is a schematic diagram of the resource level framework for resource data collection provided in an embodiment of the present invention;

[0078] Figure 4 This is a flowchart illustrating the quality analysis and data acquisition process provided in this embodiment of the invention.

[0079] Figure 5 This is a schematic diagram of the flow direction analysis process provided in an embodiment of the present invention;

[0080] Figure 6 This is a business process diagram of the resource scheduling method based on business calculation provided in the embodiments of the present invention;

[0081] Figure 7 This is a comparative diagram of the resource scheduling process provided in the embodiments of the present invention;

[0082] Figure 8 This is a schematic diagram of the structure of a resource scheduling device based on business calculation provided in an embodiment of the present invention;

[0083] Figure 9 This is another structural schematic diagram of a resource scheduling device based on business calculation provided in an embodiment of the present invention. Detailed Implementation

[0084] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] In a content delivery network, global load balancing is usually achieved through intelligent DNS. Based on the address of the LDNS configured by the customer, the cache is distributed to the nearest available cache server. Therefore, the scheduling accuracy of a CDN depends on the intelligent resolution of GSLB.

[0086] Currently, the mainstream GSLB scheduling methods are to perform forwarding scheduling through local LDNS or to perform scheduling through CNAME aliases.

[0087] Current scheduling strategies simply calculate network bandwidth and server availability, which are not suitable for the personalized scheduling of VR / AR / XR services under the new circumstances. They also lack a dynamic weighting scheduling method based on the linkage between end-to-end service quality expectation indicators and resource load expectation.

[0088] Furthermore, some CDN service providers, in order to save costs, lease IDC resources in adjacent regions (i.e., other regions) at low prices to provide services to users in their own region. This presents two problems: first, the scheduling latency between different regions is large, resulting in a reduced user experience and lower user satisfaction; second, traffic penetrates between backbone networks, wasting bandwidth and resources. This constitutes an error in operational management and is not adapted to the traffic management strategy under the new model of shifting computing power hierarchy from centralized to a three-dimensional "cloud-edge-device" approach.

[0089] The existing solutions have low accuracy and efficiency in resource scheduling, resulting in low quality of user services.

[0090] This application proposes a resource scheduling method based on business calculations, see [link to relevant documentation]. Figure 1 This is a flowchart illustrating a resource scheduling method based on business calculation provided in an embodiment of the present invention, the method including steps S1 to S5;

[0091] S1 collects scheduling resource data for traffic service scheduling in this region;

[0092] S2, perform service quality and traffic flow analysis on the scheduling resource data to determine evaluation indicators;

[0093] S3. Based on the evaluation indicators, predict the bandwidth utilization rate and service quality indicators for the next cycle, and score them respectively.

[0094] S4, determine the allocation weight of the local IDC parsing fast table based on the scoring results;

[0095] S5. Based on the allocation weight, select the local IDC parsing fast table for scheduling query to determine the resource scheduling address.

[0096] In this specific implementation, a traffic flow analysis and decision-making system is established for resource scheduling management. Leveraging the operator's control over the Local Network Domain Name System (LDNS), some scheduling authority is delegated to the LDNS. First, scheduling resource data for traffic services within the region is collected. This data includes intra-regional resource scheduling and traffic resource scheduling between the region and other regions. The scheduling data is collected via the regional network interface. The scheduling view of the region is synchronized with the operator's CDN global scheduling system (GSLB, if applicable) for this region. End-to-end service quality analysis, bandwidth utilization analysis, service analysis, and cross-regional analysis are performed to determine the current resource scheduling evaluation indicators. During evaluation and prediction, the end-to-end service quality and bandwidth utilization for the next cycle are predicted based on a preset model.

[0097] Based on the prediction results, the service quality and link bandwidth utilization of key websites are ranked, and different indicators are scored separately, and priority scheduling rights are granted.

[0098] Based on the scoring results, a weighted IDC address resolution fast table is constructed. The weighted local IDC resolution fast table is deployed in the LDNS. The data in this fast table is obtained from the traffic flow analysis and decision system. If the weight is 0, scheduling is performed from the synchronized GSLB. This achieves efficient, fast, and prioritized accurate resource scheduling locally.

[0099] This application achieves optimal scheduling through the local LDNS controlled by the operator, solving the problem of blind scheduling during resource service scheduling. By performing service calculations and traffic quality analysis, it improves the efficiency and accuracy of resource scheduling and enhances the quality of service.

[0100] In another embodiment of the present invention, when collecting scheduling resource data for traffic services in this region, the following steps are specifically performed:

[0101] See Figure 2 This is another flowchart illustrating the resource scheduling method based on business calculation provided in this embodiment of the invention.

[0102] When using a traffic flow analysis and decision-making model for resource scheduling and management, it is mainly divided into four layers: the collection layer, the analysis layer, the prediction layer, and the application layer.

[0103] The acquisition layer collects GSLB scheduling data, DNS-related IDC IP resolution data, regional network port outbound DPI traffic data, and IDC outbound DPI traffic data.

[0104] Synchronous backup of GSLB scheduling data after data acquisition;

[0105] The acquired data is then aggregated, format-converted, and grouped to synthesize an XDR call detail record.

[0106] In another embodiment provided by the present invention, step S2 is specifically implemented by performing the following steps:

[0107] The drill-down levels for different IDC customers, different services, and different domain names are determined based on the customer information, IP address database information, and domain name information in the scheduling resource data.

[0108] See Figure 3 This is a schematic diagram of the resource level framework for resource data collection provided in this embodiment of the invention. For the Internet, there are only three types of resources: customer information, IP address database, and domain name database. Customer information refers to the correspondence between IDC customers and IDC network segments. Basic annotative information about some customers can be referred to as SP (Service Provider). IP address database information refers to the correspondence between IP addresses and their locations, i.e., which operator, region, and data center an IP belongs to. Domain name information refers to the correspondence between domain names and business subcategories, divided into first-level domains and second-level domains. The correspondence only provides the first-level domain; the business subcategories to which the second-level domains belong automatically inherit from the first-level domain. Examples of first-level domains include XXX.COM, XXX.CN, XXX.ORG, etc., and second-level domains include XXX.XXX.COM, etc.

[0109] When querying domain names, all domain names should be listed. When querying business subcategories, only the subcategories with corresponding relationships should be considered. For SPs, drill-down can be performed on the first-level domain or the IP address (i.e., which IPs the SP has purchased from the IDC. You can switch between displaying IPs and network segments). For first-level domains, drill-down can be performed on second-level domains or the IP address (i.e., which IPs the first-level domain can resolve to; only the specific IPs need to be displayed, not the network segments).

[0110] By conducting specific business quality and domain name quality analysis at different drill-down levels, business quality indicators are determined. By analyzing specific business and domain name quality, relevant trends are identified, which are then sent to the prediction layer for decision-making.

[0111] By analyzing the service traffic flow rate of websites at different drill-down levels, the bandwidth utilization of IDC-related links can be determined.

[0112] The service quality indicators and the bandwidth utilization rate are used as the evaluation indicators.

[0113] During the execution of business analysis, the analysis layer completes cross-regional traffic flow analysis, business analysis, bandwidth utilization analysis, and business quality analysis for key IDC business customers, and determines the business quality indicators and the bandwidth utilization as the evaluation indicators.

[0114] In another embodiment of the present invention, the following steps are specifically performed when determining business quality indicators:

[0115] Based on the identification of different levels of domain names and IPs in the scheduling resource data, the drill-down level under different levels is determined.

[0116] See Figure 4 This is a flowchart illustrating the quality analysis and data collection process provided in this embodiment of the invention. It requires analyzing the quality of specific services and domain names. Specifically, it involves determining the service category dimensions for drilling down to the SP using a service category identification code table, performing data association, and determining the quality indicators for each category.

[0117] It should be noted that the quality indicators are different for each major category. Specific indicators can be entered into the Excel file to determine the quality indicators.

[0118] After the indicators are determined, the quality indicators of the uplink and downlink flow can be monitored during the first-level drilling, specifically including one of the following:

[0119] Using an IP address database, data association is performed on the IPs drilled down by the SP to analyze high-quality and low-quality IPs, and their respective subdomains are labeled. The HTTP response success rate, HTTP response latency, TCP establishment success rate, and TCP establishment latency of the IPs are determined as business indicators and sent to the prediction layer for decision-making.

[0120] The preset domain name database and the IP address database are used to associate the second-level domain names under the SP, analyze the high-quality domain names and low-quality domain names, and mark the SP and IP to which the second-level domain name belongs. The HTTP response success rate, HTTP response latency, TCP establishment success and TCP establishment latency of the second-level domain name are determined as business indicators and sent to the prediction layer for decision-making.

[0121] It should be noted that the quality indicators in this case can be determined by setting different indicator definitions and algorithms for data from different sources and for different business types. For example, for web page traffic business with DPI data type, the page response success rate can be set as an indicator. The probability of a user initiating a web page browsing and successfully establishing a web page connection can be defined as the indicator definition. The resulting indicator algorithm is the product of the TCP connection establishment success rate and the first GET request success rate as the page response success rate. Indicators from other data sources and business types will not be elaborated on here.

[0122] In another embodiment of the present invention, when determining the bandwidth utilization of IDC-related links, the following steps are specifically performed:

[0123] See Figure 5This is a flowchart illustrating the traffic flow analysis process provided in this embodiment of the invention. When determining bandwidth utilization through traffic flow analysis, the peak flow rate from each primary and secondary domain name of key websites to each region is considered. This includes the downlink peak flow rate and load status of the IDC SP service within its own region, other regions, the same network, and others. Specifically, this is associated with IP addresses.

[0124] Specifically, in terms of service traffic, we only focus on downlink traffic. That is, we drill down to the first-level domains of SPs in the TOP55 dimension, associate the preset IP address database with the first-level domains, and generate the peak flow rate and load of each first-level domain of important websites to each region, as well as other downlink peak flow rates of IDC SP services in other regions and other customers; that is, the other downlink peak flow rates and load of IDC SP services in this region, other regions, this network and other service providers.

[0125] Furthermore, based solely on downlink data, drill down to both uplink and downlink data. Drill down to the first-level domain of the SP in the TOP55 dimension, and further drill down to the second-level domain. Associate the IP address database with the second-level domains drilled down from the first-level domains to generate peak flow rates and load conditions for each second-level domain under each first-level domain of important websites in each region, as well as other downlink peak flow rates for IDCSP services in other regions and for other customers; that is, the downlink peak flow rates and load conditions for IDC SP services in this region, other regions, this network, and other service providers.

[0126] The bandwidth utilization of IDC-related links is determined based on traffic analysis.

[0127] In another embodiment of the present invention, the following steps are specifically performed when predicting and scoring business quality indicators:

[0128] Because regression models are ineffective for factors with complex influencing factors, especially those for which data on these factors is unavailable, and because data is a time series, it may exhibit a certain steady state or pattern over time. Using past time intervals to predict future values, while also considering cyclical factors, triple exponential smoothing is the best method. Therefore, triple exponential smoothing is applied to predict the indicator situation for the next cycle or several cycles.

[0129] The bandwidth utilization rate and service quality indicators collected during this period are calculated to determine the stability value, trend value, and seasonality value.

[0130] The average flow rate collected from the first to the i-th time within the period is the stationarity value, denoted as s. i s i =αx i / p i-k +(1-α)(s i-1 +ti-1 ).

[0131] The trend value is determined by the increasing or decreasing trend of the current value, calculated based on the difference between the i-th and (i-1)-th times, and is denoted as t. i , t i =β(s) i -s i-1 )+(1-β)t i-1 .

[0132] The seasonal value is an overall increasing or decreasing trend derived from the data collected from the first to the i-th time, denoted as p. i p i =γx i / s i +(1-γ)p i-k .

[0133] Parameter x i This represents the data collected in the i-th iteration of the original traffic data sequence. i+h This represents the predicted data value. The subscript i+h indicates the predicted value h time intervals from the i-th data collection. When h is 1, it represents the predicted value of the indicator data in the next collection. k is the number of data collections within one indicator data period. p i-k The one is with p i The periodic value corresponding to the previous period. p i-k+(hmodk) The physical meaning of is the seasonal value of the previous cycle of the sequence number i+h.

[0134] α, β, and γ are independent variables ranging from 0 to 1, primarily determining the predicted values ​​of future data. The closer they are to 1, the greater the influence of recent data values ​​on the predicted values. These independent variables can be obtained by substituting continuous actual values ​​from the collected data.

[0135] The data inputs are: associated IP bandwidth utilization and associated IP service indicators. The bandwidth utilization and service quality indicators are calculated separately to determine the stability, trend, and seasonality values ​​of the bandwidth utilization and the stability, trend, and seasonality values ​​of the service quality indicators.

[0136] The three-stage exponential smoothing method is used to calculate the stationarity, trend, and seasonality values ​​of bandwidth utilization and service quality indicators, respectively, to predict the bandwidth utilization of IP and related service indicators of IP in the next cycle.

[0137] Forecasted IP bandwidth utilization for the next cycle, and forecasted IP-related service metrics for the next cycle, x i+h =(s i +ht i )p i-k+(hmodk) .

[0138] After obtaining the predicted indicators for the next period, the bandwidth utilization rate and related service indicators of the IP for the next period are scored according to the scoring range of the indicators. Specifically:

[0139] Business forecast metrics are divided into five ranges, scored from high to low. For example, a page response success rate of 98% to 100% is scored as 100, 95% to 97.99% is scored as 80, 94.99% to 92% is scored as 60, 91.99% to 90% is scored as 40, 89.99% to 87% is scored as 20, and below 86.99% indicates severe quality deterioration and is scored as 0.

[0140] Related IP Prediction Service Quality Score: = SUM(Service Response Success Rate Score + Page Response Latency Score + ...) / Number of Service Quality Indicators / 100.

[0141] The predicted IP bandwidth utilization rate is divided into five ranges, scored from high to low. For example, the utilization rate of 0-10% is scored as 100, 11%-20% is scored as 80, 21-30% is scored as 60, 31-40% is scored as 40, 41%-50% is scored as 20, and above 51% indicates that the utilization rate is too high and is scored as 0.

[0142] Predicted bandwidth utilization score: = Utilization score / 100.

[0143] In another embodiment provided by the present invention, the scores of bandwidth utilization and service quality indicators of the same IP are weighted and summed to determine the weight of each IP of the domain name; that is, the weight of each IP of the same domain name is: =ROUND((50*relevant IP predicted bandwidth utilization score + 50*relevant IP predicted service quality score) / 10)*10 / total weight of the same domain name, 1)*100.

[0144] Summing the IP weights of all IPs under the same domain name, the total weight of the same domain name is: = SUM(weight1 of the same domain name + weight2 of the same domain name + ...), which gives the weight of each domain name, and updates the weight allocation of different domain names in the local IDC resolution table.

[0145] In a preferred embodiment, domain name AA has two corresponding A records: IP1 and IP2. IP1 has a predicted bandwidth utilization score of 0.80 and a predicted service quality score of 1; IP2 has a predicted bandwidth utilization score of 0.60 and a predicted service quality score of 0.80. Therefore: IP1 domain weight = ROUND(90 / 160, 1) * 100 = 60; IP2 domain weight = ROUND(70 / 160, 1) * 100 = 40.

[0146] Record each domain name, see Table 1, and determine the allocation weight of the local IDC DNS resolution fast table.

[0147] Table 1 Domain Name Records

[0148]

[0149] The prediction layer mainly performs triple exponential smoothing on normalized data to predict the results of service quality, bandwidth utilization, etc. in the next cycle or several cycles, thereby dynamically deriving the weight structure of the local IDC parsing fast table.

[0150] In another embodiment of the present invention, when determining the resource scheduling address in step S5, see [reference needed]. Figure 6 This is a business process diagram of the resource scheduling method based on business calculation provided in the embodiments of the present invention.

[0151] When performing a resource scheduling query, the following steps are executed:

[0152] Step 1: The DPI device collects data from router nodes and IDC nodes, collects data from the regional network egress and IDC egress to synthesize XDR call detail records, and sends them to the traffic flow analysis and decision system.

[0153] Step 2: The DNS resolution logs of the relevant IPs of the IDC in this region are sent to the traffic flow analysis and decision-making system.

[0154] Step 3: If it is a carrier CDN, then synchronize the GSLB scheduling data of this region to the traffic flow analysis and decision system.

[0155] Step 4: The traffic flow analysis and decision system calculates key website business quality, bandwidth utilization, etc. based on the above data, predicts the situation in the next period and scores it. If there is no change in the next period, the current weight scheduling allocation result is maintained. If there is a change in the next period, the current scheduling weight allocation result is updated and synchronized to the local IDC resolution fast table in this region. The local IDC resolution fast table in this region is a separate process cache area of ​​LDNS.

[0156] Step 5: The user accesses the key website, and the domain name arrives at the local LDNS.

[0157] Step 6: The local LDNS first looks up the local IDC's DNS table in this region.

[0158] Step 7: If the local IDC resolves the fast table and finds a result with a weight that is not 0, then the result is fed back to LDNS.

[0159] Step 8: The local LDNS sends the resolution result back to the user.

[0160] Step 9: If the LDNS does not find the record in the local IDC fast table of this region, or the weight is 0 factors or there is no record, then a full cache lookup is performed. If the record is found, the resolution result is returned to the user; if the record is not found, a recursive lookup is performed in the authoritative DNS.

[0161] Step 10: Authorize the DNS to provide the address of this CDN domain alias.

[0162] Step 11: The local LDNS is used to resolve and look up the GSLB scheduling data.

[0163] Step 12: GSLB returns the nearest resource address to LDNS, and then proceeds to step 7 to return the address to the user.

[0164] In practice, scheduling queries should be performed first in the local IDC parsing fast table.

[0165] Websites outside the operator's CDN distribution range but with resources within the operator's local IDC are defined. To prevent incorrect scheduling, an error correction mechanism is established. For example, analyzing the inbound and outbound traffic of various services during peak evening hours, the data collection points are the Internet regional network port outbound DPI traffic and the IDC outbound and outbound traffic. This is used to analyze situations where the top 55 websites are incorrectly scheduled across regions when resources exist in the IDC. By establishing an error correction mechanism, the service quality of IDC-related IPs is statistically analyzed. If resource utilization is low and service quality is high, a score is assigned based on the prediction results using triple exponential smoothing, and weights are allocated. A local IDC DNS fast table is established, recording the A address and weight corresponding to the website domain name. This directly completes high-traffic resource scheduling, shortens latency, and prevents cross-regional scheduling. If service quality drops to a certain percentage, the weight ranking results determine whether to use GSLB for scheduling.

[0166] Second, if resource scheduling is completed within the operator's CDN, an automatic matching weighted scheduling method is constructed based on IDC resource conditions, load conditions, and service quality conditions. A three-stage exponential smoothing method is used to predict forward-looking indicators, dynamically adjust weight ratios, and periodically update the data. Data is injected into the parsed fast table and synchronized with GSLB data. If the data is consistent with GSLB data, scheduling is based on the parsed fast table data. If inconsistent and the weight has not dropped to 0, scheduling is based on the fast table and synchronized to GSLB. If the weight drops to 0, scheduling is based on the GSLB result.

[0167] The traffic flow analysis and decision-making system dynamically adjusts the distribution strategy for user requests based on calculated weights. See also... Figure 7 This is a comparative diagram of the resource scheduling process provided in the embodiments of the present invention.

[0168] This case dynamically adjusts the user request distribution strategy based on calculated weights. For example, for nodes with heavy loads and declining service quality, their request distribution weight is reduced, while for nodes with light loads and good service quality, their request distribution weight is increased.

[0169] By leveraging the LDNS controlled by the operator for scheduling, a local IDC resolution fast table is deployed on the LDNS side. If it is the operator of the same network, it is synchronized with GSLB. This solves the problem of incorrect scheduling across regions when some local resources exist and are available in CDN operation, reduces internet latency, and improves customer experience.

[0170] This application addresses the content scheduling issues in current CDNs. Some DNS providers do not support EDNS, meaning the source address of a user's website access is not resolved and sent to GSLB. Some CDN providers implement EDNS, which can obtain the user's source IP and schedule resources based on proximity; however, after the TTL expires, each access requires access to the authorization server for scheduling. Some CDN providers implementing EDNS, in order to reduce IDC rental costs or due to misconfigurations in the source address and proximity resource matching table, incorrectly schedule resources from adjacent regions to users in their own region, directly prioritizing local resources. This application addresses the issues of deviations between actual scheduling results and preset weight policies in GSLB's global scheduling, and the potential for large traffic fluctuations due to frequent address switching for LDNS requests when cache expires. It proposes predictive weighted scheduling.

[0171] This invention requires no modification to recursive DNS or GSLB, and no additional fields are needed. Weight calculation is performed solely within the traffic flow analysis and decision-making system, and the LDNS undergoes a fast table modification, thus improving resolution speed. It avoids the high load associated with frequent weight scheduling by GSLB.

[0172] See Figure 8 This is a schematic diagram of a resource scheduling device based on business calculation provided in an embodiment of the present invention. The device includes:

[0173] The data acquisition module is used to collect scheduling resource data for traffic service scheduling in this region;

[0174] The analysis module is used to perform service quality and traffic flow analysis on the scheduling resource data and determine evaluation indicators;

[0175] The prediction module is used to predict the bandwidth utilization rate and service quality indicators for the next period based on the evaluation indicators, and to score them respectively.

[0176] The weighting module is used to determine the allocation weight of the local IDC parsing fast table based on the scoring results;

[0177] The scheduling module is used to select the local IDC parsing fast table for scheduling based on the allocation weight, and determine the resource scheduling address.

[0178] It should be noted that the resource scheduling device based on business calculation provided in this embodiment can execute all the steps and functions of the resource scheduling method based on business calculation provided in any of the above embodiments. The specific functions of the device will not be described in detail here.

[0179] See Figure 9 This is another structural schematic diagram of a resource scheduling device based on business calculation provided in an embodiment of the present invention. The resource scheduling device based on business calculation includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a resource scheduling program based on business calculation. When the processor executes the computer program, it implements the steps in the various embodiments of the resource scheduling method based on business calculation described above, for example... Figure 1 The steps S1 to S5 are shown. Alternatively, when the processor executes the computer program, it implements the functions of each module in the above-described device embodiments.

[0180] For example, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the resource scheduling device based on business calculation. For example, the computer program can be divided into several modules, the specific functions of which have been described in detail in the resource scheduling method based on business calculation provided in any of the above embodiments; therefore, the specific functions of the device will not be repeated here.

[0181] The resource scheduling device based on business calculation can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. This resource scheduling device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the schematic diagram is merely an example of a resource scheduling device based on business calculation and does not constitute a limitation on such a device. It may include more or fewer components than illustrated, or combine certain components, or use different components. For example, the resource scheduling device based on business calculation may also include input / output devices, network access devices, buses, etc.

[0182] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the service-based resource scheduling device, connecting all parts of the device via various interfaces and lines.

[0183] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the resource scheduling device based on business calculation by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0184] The module integrated into the resource scheduling device based on business calculations, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0185] This invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method described in any of the above embodiments.

[0186] It should be noted that the computer program product provided in this embodiment can execute all the steps and functions of a resource scheduling method based on business calculation provided in any of the above embodiments, and the specific functions of the device will not be described in detail here.

[0187] It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications are also considered to be within the scope of protection of this invention.

Claims

1. A resource scheduling method based on business calculations, characterized in that, The method includes: Collect scheduling resource data for traffic service scheduling in this region; The scheduling resource data is analyzed for service quality and traffic flow to determine evaluation indicators; Based on the aforementioned evaluation indicators, predict the bandwidth utilization and service quality indicators for the next cycle, and score them accordingly. The allocation weight of the local IDC parsing fast table is determined based on the scoring results; Based on the allocation weight, the local IDC parsing fast table is selected for scheduling query to determine the resource scheduling address; The scheduling resource data is analyzed for service quality and traffic flow to determine evaluation indicators, including: The drill-down levels for different IDC customers, different services, and different domain names are determined based on the customer information, IP address database information, and domain name information in the scheduling resource data. By conducting specific business quality and domain name quality analyses at different drilling levels, business quality indicators are determined. By analyzing the service traffic flow rate of websites at different drill-down levels, the bandwidth utilization of IDC-related links can be determined. The service quality indicators and the bandwidth utilization rate are used as the evaluation indicators.

2. The resource scheduling method based on business calculation according to claim 1, characterized in that, Collect scheduling resource data for traffic service scheduling in this region, including: Obtain GSLB scheduling data, DNS-related IDC IP resolution data, regional network port outbound DPI traffic data, and IDC outbound DPI traffic data for traffic services in this region; Perform synchronous backup of the GSLB scheduling data; The acquired data is format-converted and grouped to synthesize an XDR call detail record.

3. The resource scheduling method based on business calculation according to claim 1, characterized in that, Define business quality indicators, including: The system associates the pre-defined IP address database with the IPs drilled down by the SPs in the business category, analyzes and identifies high-quality and low-quality IPs, marks their respective subdomains, and determines the IPs' HTTP response success rate, HTTP response latency, TCP establishment success rate, and TCP establishment latency as business metrics. The preset domain name database, the IP address database, and the second-level domain names drilled down by the SP of the business category are correlated to analyze high-quality domain names and low-quality domain names. The SP and IP of the second-level domain name are labeled, and the HTTP response success rate, HTTP response latency, TCP establishment success rate, and TCP establishment latency of the second-level domain name are determined as business indicators.

4. The resource scheduling method based on business calculation according to claim 1, characterized in that, Determine the bandwidth utilization of IDC-related links, including: The system associates the preset IP address database with the first-level domains drilled down by the SP for major business categories to generate peak flow rates and load conditions for each first-level domain of important websites to each region, as well as other downlink peak flow rates for IDC SP services to other regions and other customers. The IP address database and the subdomains drilled down from the first-level domain are associated to generate peak flow rates and load conditions for each subdomain under each first-level domain of important websites in each region, as well as other downlink peak flow rates for IDC SP services in other regions and for other customers; The bandwidth utilization of IDC-related links is determined based on traffic analysis.

5. The resource scheduling method based on business calculation according to claim 1, characterized in that, Based on the aforementioned evaluation metrics, predict the bandwidth utilization and service quality metrics for the next cycle, and score them accordingly, including: The bandwidth utilization rate and service quality indicators collected during this period are calculated to determine the stability value, trend value, and seasonality value. The three-stage exponential smoothing method is used to calculate the stationarity, trend, and seasonality values ​​of bandwidth utilization and service quality indicators, respectively, to predict the bandwidth utilization of IP and related service indicators of IP in the next cycle. The bandwidth utilization rate and related business indicators of the IP in the next cycle are scored according to the pre-defined scoring intervals.

6. The resource scheduling method based on business calculation according to claim 1, characterized in that, The allocation weight of the local IDC parsing fast table is determined based on the scoring results, including: The weights of each IP address are determined by weighted summation of the scores for bandwidth utilization and service quality indicators for the same IP address. Sum the IP weights of all IPs under the same domain name and update the assigned weights of different domains in the local IDC resolution table.

7. The resource scheduling method based on business calculation according to claim 1, characterized in that, Based on the allocated weights, the local IDC parsing fast table is selected for scheduling queries to determine the resource scheduling address, including: The lookup is performed based on the local IDC parsing fast table; When a resource address is found and the corresponding allocation weight is not 0, it is fed back to LDNS for resource scheduling. When no resource address is found, or the allocation weight corresponding to the found resource address is 0, the GSLB scheduling data in the scheduling resource data is parsed and searched, and the latest resource address found is fed back to LDNS for resource scheduling.

8. A resource scheduling device based on business calculation, characterized in that, The device includes: The data acquisition module is used to collect scheduling resource data for traffic service scheduling in this region; The analysis module is used to perform service quality and traffic flow analysis on the scheduling resource data and determine evaluation indicators; The prediction module is used to predict the bandwidth utilization rate and service quality indicators for the next period based on the evaluation indicators, and to score them respectively. The weighting module is used to determine the allocation weight of the local IDC parsing fast table based on the scoring results; The scheduling module is used to select the local IDC parsing fast table for scheduling based on the allocation weight, and determine the resource scheduling address.

9. A resource scheduling device based on business calculation, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements a resource scheduling method based on business calculation as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform a resource scheduling method based on business calculation as described in any one of claims 1 to 7.

11. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method described in any one of claims 1 to 7.

Images