Method for realizing efficient scheduling and shunting of ACS system based on NGINX
The dynamic routing table and busy factor selection are achieved through the combination of Redis+NGINX+Lua, which solves the balance and security of the scheduling method in the ACS system, improves the response speed and system resource utilization of ACS and CPE sessions, and avoids request skew , realizes efficient scheduling and scalability of the ACS system.
Patent Information
- Application Number
- CN202410968233.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-15
AI Technical Summary
In terms of scheduling methods, existing ACS systems are difficult to balance session security, response speed and server resource utilization, especially when the number of CPEs increases, it is easy to cause request tilt and resource waste.
The combination of Redis+NGINX+Lua is used to realize dynamic routing tables and busy factor selection, ensuring the security and response speed of ACS and CPE session connections, and improving system scalability by dynamically increasing and decreasing CWMP instances.
It realizes efficient scheduling of ACS and CPE sessions, prevents request skew, improves system utilization and response speed, and ensures the security of ACS and CPE sessions and the effective utilization of resources.
Smart Images

Figure CN118972338A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data decision making, and in particular to an efficient scheduling and diversion method for an ACS system based on NGINX. Background Art
[0002] TR069 is a data communication protocol for network devices with a wide range of uses.
[0003] Currently, TR protocol is used for communication between CPE and automatic configuration server ACS. This protocol provides a set of mechanisms, including reliable automatic configuration of CPE, and merges some other CPE management functions into the same common platform framework.
[0004] In the ACS system, the number of managed CPEs, session response speed, and session connection security are three important indicators. Therefore, the system needs to be supported by a distributed architecture, and efficient scheduling becomes particularly important. In addition, due to business needs, the ACS system must be able to directly configure CPE remotely. How to ensure the stability of the session connection between ACS, CWMP, and CPE is also a pain point in the scheduling process.
[0005] Four common scheduling methods:
[0006] Round_Robin: round-robin, one round after another
[0007] Weight_Round_Robin: Weighted round-robin, one round at a time, but with different weight ratios
[0008] IP_Hash: Requests from the same IP address will be dispatched to the same server
[0009] Minimum connections: The request will be dispatched to the server with the smallest number of connections.
[0010] In actual work, there are two ways of session connection:
[0011] (1) CPE initiates a connection:
[0012] CPE can use the pre-defined unified domain name address to establish a connection with the ACS system at any time. CPE must initiate a connection with the ACS and send the Inform RPC method, including:
[0013] a) When the CPE is initially installed, it initiates the first access to the network and initiates the connection;
[0014] b) Power on or restart;
[0015] c) Heartbeat, each notification cycle interval;
[0016] d) When instructed to execute by the optical ScheduleInform method;
[0017] e) The CPE receives a valid link request from the ACS;
[0018] f) The URL of ACS is changed;
[0019] g) A parameter that requires an Inform to be issued when it changes has changed;
[0020] h) A parameter marked as an active notification by the ACS through the SetParameterAtributes method changes due to an external reason. The change of the parameter by the ACS itself through SetParameterValues must not cause a new session to be initiated. If a parameter is modified multiple times before the CPE can successfully initiate a session to execute this notification, the CPE only needs to report the notification once. If a parameter changes due to internal factors during the course of a session, a new session will be initiated and established after the end of the current session (it must not affect the current session);
[0021] i) In the event of an abnormal termination, re-establishment will be initiated according to the session re-establishment procedure.
[0022] PS: When there are no messages waiting to be processed by the CPE or ACS, the CPE must not continue to maintain the link to the ACS and the session is terminated.
[0023] (2) ACS can use the connection request mechanism to request CPE to initiate a connection to ACS at any time. CPE needs to support this mechanism, and ACS also needs to support this mechanism. This mechanism includes:
[0024] a) The connection request must be an HTTP GET to a URL specified by the CPE, and the URL value is a valid read-only parameter on the CPE. The channel value of this URL should be randomly generated by the CPE, so it is unique to a particular CPE;
[0025] b) The connection request must use HTTP, not HTTPS;
[0026] c) The HTTP GET of the connection request should not contain any data. The CPE should ignore any data that may exist in it;
[0027] d) The CPE must authenticate the ACS using Digest-Authentication before execution;
[0028] e) The CPE MUST allow connection requests from any source address that has the correct authentication parameters for the target CPE;
[0029] f) The CPE responds to a properly authenticated connection request with an HTTP status code of "200 (Ok)" or "204 (Not Content)". The CPE MUST send this response immediately after successful authentication and before initiating the session. At the same time, the message body length in the HTTP response MUST be 0;
[0030] g) To prevent the CPE from being attacked by services, the CPE should limit the number of connection requests it can accept within a certain period of time. If the CPE rejects a connection request for this reason, the CPE must respond to the connection request with an HTTP 503 (Invalid Server). In this case, the CPE should not include the HHT Retry-After header in the response message;
[0031] h) If the CPE successfully authenticates a connection request and responds as described above, and it is not already in a session, it MUST attempt to establish a session using the ACS address contained in the "6CONNECTIONREQUEST" EventCode in the inform within 30 seconds of sending the response;
[0032] i) If the ACS receives a successful response to a connection request, but the CPE is still unable to successfully establish a connection based on the ACS address contained in the "6CONNECTION REQUEST" EventCode in the inform within 30 seconds, the ACS can re-initiate the connection request to the CPE;
[0033] j) If the CPE receives one or more other successfully authenticated connection requests after successfully authenticating and responding to a connection request and before establishing a session with the ACS, the CPE MUST return a successful response to each connection request, but MUST NOT establish additional sessions for those other connection requests, regardless of how many connection requests were received during this time.
[0034] k) If the CPE receives one or more requests while it is already in a session with the ACS, it MUST NOT terminate the current session prematurely, but MUST either respond with an HTTP 503 status code, or, after the current session ends, initiate a new session with the address contained in the "6CONNECTION REQUEST" EventCode in the inform;
[0035] l) The CPE MUST NOT reject a fully authenticated connection request for reasons other than those described above. If the CPE rejects the connection request for one of the reasons described above, it MUST NOT initiate a session to the ACS for the connection request.
[0036] Existing methods 1, 2, 4, the most common scheduling methods include:
[0037] In the current scheduling scenario, the round-robin mechanism can ensure that each CWMP receives a very balanced number of requests, but it can easily lead to insecure sessions between the CPE and the ACS.
[0038] If the ACS initiates a session request to the CPE through the CWMP-1 instance, the CPE will respond and try to establish a session with the ACS. If the current scheduling is a round-robin mechanism, the CPE's response will be round-robin to the CWMP-2 instance. In this way, the CWMP-1 instance will not receive the CPE's response, which will eventually lead to an insecure session.
[0039] The third existing method is a method based on IP_Hash scheduling. If the request is from the same IP client, IP_Hash will schedule the request origin to a unified server. IP_Hash is detected based on the first few digits of the IP address.
[0040] In the IP_Hash scheduling scenario, the session security between ACS and CPE can be well guaranteed. However, as the number of managed CPEs increases, request skew is prone to occur, and all CPE requests are concentrated in a few CWMP service instances, while other CWMP service instances are idle. This ultimately leads to slow response and low resource utilization between ACS and CPE sessions.
[0041] Through the above analysis, the existing scheduling method cannot meet the needs of the ACS system well. The existing problems and defects are: it is difficult to simultaneously meet the balanced performance and response speed of the scheduling method, the security of the ACS and CPE sessions, and the server utilization.
[0042] The difficulty of solving the above problems and defects is:
[0043] To ensure the security of the ACS and CPE sessions, in a distributed architecture, the ACS and CPE sessions are always executed through the same CWMP instance over a period of time. However, the executed CWMP is not fixed, but should be dynamic, which is the best choice.
[0044] To prevent request skew, try to avoid a CWMP instance from executing too many session requests while ensuring that the above problems are solved.
[0045] The significance of solving the above problems and defects is:
[0046] The present invention will use the combination of Redis+NGINX+Lua to implement a dynamic routing table to solve the scheduling problem of ACS and CPE session requests and the data distribution isolation function;
[0047] The present invention utilizes the busy factor to solve the session request tilt problem. Summary of the invention
[0048] The technical problem to be solved by the present invention is to overcome the above technical defects and provide an efficient scheduling method and system based on NGINX to meet the TR protocol interaction between ACS and CPE. The system management includes a combination of one or more of the following: information acquisition, parameter configuration, performance monitoring, performance alarm, and fault diagnosis.
[0049] The present invention provides a temporary routing table, which can improve the session response speed and system utilization as much as possible under the premise of ensuring the security of the ACS and CPE session connection. The present invention provides a busy factor to prevent the occurrence of connection tilt. The present invention provides a dynamic increase and decrease of CWMP instances to improve the scalability of the system.
[0050] The present invention is implemented as follows: a scheduling method for the TR communication protocol between the ACS and the CPE based on NGINX, including: security access verification, selecting CWMP service based on busy factor, preventing request connection skew, and a connection request method based on dynamic routing table scheduling.
[0051] Furthermore, the scheduling method of the TR communication protocol between the ACS and the CPE based on NGINX, when the CPE initiates a connection, specifically includes:
[0052] Step 1: When the CPE initiates a connection request to the ACS, it first resolves the unified URL configured in the CPE through DNS to obtain the IP, and then initiates a connection request to the IP.
[0053] Step 2: NGINX on the server receives and starts processing the request initiated by the CPE.
[0054] Step 3: Dispatch obtains the IP address of the requested CPE and checks whether it is the first access. If it is the first access, the backend verifies whether the CPE model and brand are managed.
[0055] Step 4: When the verification is passed or it is not the first access, Dispatch will request the routing table. If the CPE IP exists in the routing table, it will directly respond through the CWMP found in the temporary dynamic routing table. If it does not exist, it will select the optimal CWMP to respond based on the busy factor.
[0056] Step 5: After the response is completed, write or update the new routing information into the temporary dynamic routing table and set the validity period.
[0057] Furthermore, when ACS initiates a connection, there are usually three situations in the business scenario. One is daily tasks, including CPE registration, upgrade and other operations. The tasks are established by ACS and CWMP performs communication with CPE. The second situation is CPE fault handling. Generally, ACS needs to operate the specified CPE, including:
[0058] Step 1: When ACS wants to operate a specified CPE, ACS first checks the routing table. If the accessed CPE exists, it directly obtains the CWMP IP from the routing table. If the accessed CPE does not exist, it searches for the CWMP IP through the busy factor.
[0059] Step 2: After finding the IP address of CWMP, ACS modifies the request IP address, disguises itself as the target CPE, and initiates a request to CWMP.
[0060] Step 3: After receiving the request from ACS, CWMP will initiate a session connection to the target CPE and send an operation message.
[0061] Step 4: After receiving the operation instruction from CWMP, CPE performs the instruction operation and reports the result to CWMP according to the above CPE initiating connection steps 1-3;
[0062] Step 5: After receiving the result, CWMP replies to the ACS request response.
[0063] refer to Figure 2 , ACS operates the link graph of the specified CPE.
[0064] The third case is that ACS periodically obtains specific parameter indicator data, which is called collection. At the same time, this part of TR interaction will be distinguished from other connection requests according to Dispatch and processed independently. It includes:
[0065] Step 1: Once the cycle is reached, the ACS proactively initiates a request to the CPE to obtain the specified performance indicator parameter process, which is similar to steps 1-3 in the second case, and is marked in REDIS.
[0066] Step 2: After receiving the operation instruction from CWMP, CPE responds to the instruction and reports the result to CWMP.
[0067] Step 3: After receiving the result, CWMP will forward it to the ACS server that processes the collected data separately according to Dispatch.
[0068] Step 4: ACS receives the response, removes the REDIS mark, and stores the performance indicator data.
[0069] Further, the dynamic temporary routing table in step 4 of the CPE initiating the connection is stored in the memory for easy reading; the information of the routing table includes: the IP of the CPE, the IP of the responding CWMP, the IP of the standby CWMP, and the valid duration of the data; the responding main CWMP is determined by the busy factor, and the standby CWMP is selected from other racks; the routing data that has not been read after the valid duration has expired will be discarded; in addition, each time the routing data is read, the valid duration of the data will be updated to ensure the security of the connection session between the ACS and the CPE;
[0070] Furthermore, the busy factor, the busy factor mentioned above is determined by the number of connections of each CWMP server and the tilt threshold, and is used for link selection of CWMP when the CPE initiates connection establishment.
[0071] Another object of the present invention is to provide an ACS system for an efficient TR protocol scheduling method implemented based on NGINX, including:
[0072] ACS: A user interaction server that provides functions such as operation, monitoring, and task delivery for managed CPEs.
[0073] CWMP: In ACS, the component that actually interacts with CPE devices.
[0074] CPE: Customer Premises Equipment, or managed equipment.
[0075] DISPATCH: Dispatching module, a custom NGINX module written in LUA.
[0076] DNS: Domain Name Server. It helps resolve URL parameters when ACS and CPE use URL addresses to identify and access each other.
[0077] DHCP: Dynamic Host Configuration Protocol server, which assigns IP addresses to ACS and CPE.
[0078] Redis: cache database.
[0079] refer to Figure 3 , ACS architecture diagram
[0080] Another object of the present invention is to provide a distributed system to improve the limit on the number of managed CPEs. Due to resource limitations and the need to ensure response speed, a CWMP can usually support 50,000 to 80,000 CPE services. A city may have hundreds of thousands to millions of CPEs, which generally require multiple CWMPs to support. A good scheduling method is particularly critical in a distributed architecture.
[0081] The advantages of the present invention compared with the prior art are:
[0082] 1. The scheduling method of the present invention can increase the upper limit of the number of CPEs managed by the ACS system, facilitate the unified management of CPEs, and reduce labor costs; in the scheduling process, the response is fast, the performance is better, and the session connection security is guaranteed; in the process of network transmission, the routing line is optimized and the resource utilization rate is improved; the ACS can directly remotely control the CPE without manual on-site visits, and CPE user failures can be solved. BRIEF DESCRIPTION OF THE DRAWINGS
[0083] Figure 1 It is the component architecture of the TR protocol.
[0084] Figure 2 It is a link diagram of the ACS operation specified CPE.
[0085] Figure 3 This is the system architecture diagram of the entire ACS.
[0086] Figure 4 The figure is a flow chart of the interactive connection between ACS and CPE.
[0087] Figure 5 The figure is a schematic diagram of the startup information reporting process between CPE and ACS. DETAILED DESCRIPTION
[0088] In order to make the contents of the present invention more clearly understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.
[0089] In view of the problems existing in the prior art, the present invention provides an efficient scheduling method and system based on NGINX to meet the requirements of TR protocol interaction between ACS and CPE. The present invention is described in detail below with reference to examples.
[0090] In order to realize the management and operation and maintenance business of three-in-one terminals (ONU, EOC, CATA) for the company, the present invention mainly develops the basic functions of the traditional ACS system, targets the management functions of the three-in-one terminals, and expands and strengthens the remote fault diagnosis direction. The present invention is mainly aimed at ONU terminals, which support the TR069 protocol. At present, the company has about one million ONUs in the existing network. Due to the limitations of performance requirements and server resources, the number of ONUs supported by a CWMP service is about 50,000 to 80,000. If the ACS service wants to support the ONU equipment of the entire network, a distributed architecture must be adopted.
[0091] In summary, the present invention is mainly aimed at the scheduling method of TR protocol connection establishment when ACS and ONU are interacting with each other. The commonly used scheduling methods are mainly round-robin, IP_hash, and minimum link. Although round-robin can ensure load balancing, it cannot ensure the security of the session between ACS and ONU. This is related to the TR protocol mechanism. For example, when ACS wants to configure and issue tasks to an ONU, ACS will first issue instructions for the configuration task to one of the CWMPs. The CWMP is named CWMP-1. At this time, CWMP-1 will directly initiate the connection establishment to the target ONU and tell it to upgrade the configuration. After receiving the request from CWMP-1, ONU will reply with an HTTP status code, such as 200. After receiving the 200 status code, CWMP-1 will start waiting. At the same time, the ONU will start to upgrade its configuration. After the upgrade is completed, it will report through the unified URL configured by itself. After the URL is parsed, when NGINX is scheduled, due to the round-robin method, it cannot guarantee that the ONU's reply connection request will be scheduled to CWMP-1. It is very likely to be scheduled to other CWMP instances, resulting in session insecurity; IP_hash, as the name suggests, means that all requests for the same IP will be scheduled to the same responder / server. Although IP_hash can ensure the session security between ACS and ONU, it may cause request skew, resulting in performance and resource waste. For example, within a certain period of time, a large number of ONUs with the same first few IPs report, and it is easy for individual CWMPs to have much more request connections than other CWMPs; minimum link, each time it is scheduled to the CWMP with the least current connections. The scheduling method of the minimum link has the same problem as round-robin. Although it can ensure load balancing, it cannot guarantee the session security between ACS and ONU.
[0092] Due to the above reasons, in actual applications, the commonly used scheduling methods cannot meet business requirements. Therefore, an efficient scheduling method based on NGINX implementation that satisfies the TR protocol interaction between ACS and CPE has been developed. Specifically, it includes: a temporary dynamic routing table that can help ONU quickly find the CWMP service instance that responds correctly; a busy factor that can help ONU find the optimal request link path and prevent request skew problems.
[0093] In addition to being applicable to ACS systems, the present invention can also meet the needs of other systems that implement forwarding scheduling based on NGINX, such as gateway control systems.
[0094] In a general embodiment of the present invention, an efficient scheduling method based on NGINX to meet the requirements of TR protocol interaction between ACS and CPE specifically includes:
[0095] Prerequisite: When the CPE goes online, a URL or domain name pointing to NGINX will be set. If it is a domain name, DNS will automatically resolve it into a real URL pointing to NGINX.
[0096] Step 1: When NGINX receives a request, the Dispatch module written in Lua will first check whether the requested device is accessed for the first time. If it is the first time, the device security verification will be performed. Otherwise, the request will be released. At the same time, Lua will also distinguish whether the current connection belongs to the periodic data acquisition function (collection) and forward the CWMP service group that does not support the function.
[0097] Step 2: After the request is verified, Dispatch will check whether the requesting device exists in the routing table in memory. If so, it will be scheduled according to the link in the routing table. Otherwise, it will select the link for scheduling based on the busy factor and forward it to the corresponding CWMP service group based on the judgment result of Lua. If the request verification fails, the request will not be processed directly, that is, the request will be interrupted.
[0098] Step 3: When the scheduling request is successful, the link information is written / updated into the routing table in the memory and the validity period is set;
[0099] Step 4: Some unimportant device reported information can be filtered to reduce network and storage pressure. For important information, important data reported by the device can be saved in real time.
[0100] Step 5: ACS will respond with the corresponding message format data according to the received SOAP data message to maintain the entire long connection request.
[0101] Step 6: ACS actively initiates a request to CPE to obtain the desired performance indicator results; ACS directly connects to the device based on the CPE-URL automatically reported by CPE, wakes up the device's CWMP reporting mechanism, and performs handshakes and exchange information on this long link.
[0102] In a general embodiment of the present invention, the busy factor in step 2 comprises:
[0103] The current number of connections and the skew threshold of the responder / server.
[0104] In a general embodiment of the present invention, the routing table of step three includes:
[0105] The temporary network link relationship requested by the device is recorded, and a backup scheduling link is set; if the link is not used for a certain period of time, the link will be abandoned. The next time the device requests, a new request link will be preferentially called based on the busy factor.
[0106] In a general embodiment of the present invention, the present invention provides an efficient scheduling method based on NGINX to meet the needs of ACS and CPE for TR protocol interaction, including:
[0107] ACS: A user interaction server that provides functions such as operation, monitoring, and task delivery for managed CPEs.
[0108] CWMP: In ACS, the component that actually interacts with CPE devices.
[0109] CPE: Customer Premises Equipment, or managed equipment.
[0110] DISPATCH: Dispatching module, a custom NGINX module written in LUA.
[0111] DNS: Domain Name Server. It helps resolve URL parameters when ACS and CPE use URL addresses to identify and access each other.
[0112] DHCP: Dynamic Host Configuration Protocol server, which assigns IP addresses to ACS and CPE.
[0113] Redis: cache database.
[0114] The technical solution of the present invention is further described below in conjunction with specific embodiments.
[0115] Example
[0116] In one embodiment of the present invention, during the request process, the terminal device / other system requests through a unified domain name, which reaches NGINX after DNS resolution; Dispatch in NGINX will first verify the device, mainly to determine whether it is the first access and whether it has passed the security verification according to the rules; after verification, it will first check the routing table cached in Redis to see if there is routing information. If it exists, it will be scheduled according to the routing information; otherwise, it will be scheduled based on the busy factor; after successful scheduling, the network link information requested by the device will be stored / updated in the cached routing table, and a timeout period will be set; within the specified time, if the routing data is not used, the scheduling link will be considered an inactive link, and the link will be abandoned to prevent data redundancy and request tilt.
[0117] The present invention is further described below by comparing the traditional scheduling with the solution of the present invention.
[0118] The traditional scheduling method is as follows:
[0119] 1)Round_Robin: round robin, one by one.
[0120] 2)Weight_Round_Robin: Weighted round-robin, one round after another, but with different weight ratios.
[0121] 3) IP_Hash: Requests from the same IP address will be dispatched to the same server.
[0122] 4) Minimum connections: The request will be dispatched to the server with the smallest number of connections.
[0123] The scheduling method provided by the present invention comprises:
[0124] 1) Security verification will be performed on the first visit to prevent malicious requests.
[0125] 2) Usually, scheduling is performed through a temporary dynamic routing table in the cache. The routing table contains two scheduling link data: primary and backup (the backup link usually selects another server). Data that has not been obtained for a certain period of time will be marked as inactive and cleared.
[0126] 3) If there is no device information in the cached routing table, the device will select the requested link based on the busy factor (determined by the current number of connections on the responding end and the skew threshold) and store the link information in the routing table.
[0127] 4) After dynamically increasing or decreasing the response end, the scheduling method can dynamically implement rebalance.
[0128] Through the above technical comparison:
[0129] During the scheduling request process, the existence of the routing table is mainly to ensure the security of the session connection between the device / client and the ACS / server, improve the scheduling speed of NGINX, and reduce the number of times the busy factor in the database is read.
[0130] It should be noted that the embodiments of the present invention can be implemented by combining software and hardware. The hardware part can be implemented using NGINX+REDIS; the software part can be implemented using code, with the NGINX module written in LUA executing the scheduling method, and the system and external interface written in JAVA. A person of ordinary skill in the art can understand that the above-mentioned devices and methods can be implemented using a combination of software and hardware.
[0131] The present invention and its implementation methods are described above, and such description is not restrictive. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design structures and embodiments similar to the technical solution without creativity, which should all fall within the protection scope of the present invention.
Claims
1. An efficient scheduling and diversion method for ACS system based on NGINX, characterized in that: The following steps are involved: Based on the scheduling method of the session between ACS and CPE, a temporary dynamic routing table is created for scheduling, and the active and standby nodes, CPE report verification, flexible increase and decrease of CWMP instances, and session skew are set.
2. According to claim 1, a method for implementing efficient scheduling and diversion of ACS system based on NGINX, characterized in that: The situations where the CPE actively initiates a session include: power-on, restart, weekly report, configuration change (external reason change); the scheduling method based on CPE initiating a session specifically includes: Step 1: The CPE is configured with the URL of the ACS at the factory, or the URL is configured uniformly by the user. The hostname component in the URL is used by DNS to find the IP address of the ACS to initiate a session. Step 2: The ACS server determines whether the CPE initiates link establishment for the first time. At the same time, the server performs management verification on the CPE that initiates link establishment for the first time. If the management verification is passed, access is allowed; otherwise, access is not allowed. Step 3: If the CPE initiates link building for the first time and is managed successfully, the Dispatch scheduler will check the busy factors of all CWMP service instances and select the CWMP service with the smallest busy factor to respond. If the CPE does not initiate link building for the first time, the Dispatch scheduler will first check the routing table in the cache and select a CWMP instance to respond based on the routing table information. If the CPE is not in the routing table, the Dispatch scheduler will check the busy factors of all CWMP service instances and select the CWMP service with the smallest busy factor to respond. Step 4: When the CWMP service responds, the information of CPE, active and standby CWMP services, etc. is written into the routing table of the dynamic cache, and the validity period is set.
3. According to claim 1, a method for implementing efficient scheduling and diversion of ACS system based on NGINX, characterized in that: The business scenarios where the ACS actively initiates an asynchronous session include remotely changing CPE configuration and remotely resetting user WIFI; The scheduling method for initiating an asynchronous session based on ACS specifically includes: Step 1: ACS can only initiate a connection session with an online CPE, and must pass 401 authentication. Step 2: ACS checks the routing table in the cache through the Dispatch scheduler first, and selects a CWMP instance based on the routing table information. If the CPE is not in the routing table, it checks the busy factors of all CWMP service instances and selects the CWMP service with the smallest busy factor. ACS will pretend to be CPE to initiate a connection session to the selected CWMP instance and update the routing table information. Step 3: After receiving the connection session from ACS disguised as CPE, CWMP will directly initiate an asynchronous session to CPE. All subsequent communications are conducted between CPE and CWMP service. ACS will not directly establish a connection session with ONU. Step 4: The CWMP service encapsulates the CPE reply information, returns the response to the ACS, and ends the session.
4. According to claim 1, a method for implementing efficient scheduling and diversion of ACS system based on NGINX, characterized in that: The data of the dynamic cache routing table is stored in the memory, with fast reading and writing. Each CPE corresponds to multiple CWMP service instances to ensure high availability. The routing table changes in real time, and the CWMP service instance is selected according to the busy factor. The timeliness of each data is controlled by timeout. Within a certain period of time, the same CWMP conducts asynchronous conversation with the CPE to ensure the security of the conversation between ACS, CWMP, and CPE.
5. According to claim 1, a method for implementing efficient scheduling and diversion of ACS system based on NGINX, characterized in that: In the process of limiting the occurrence of session tilt, when the proportion of the number of connections of a CWMP instance to the total number of connections is greater than a threshold, scheduling of new CPE connection sessions will be stopped.
6. According to claim 1, a method for implementing efficient scheduling and traffic diversion of an ACS system based on NGINX, characterized in that: During the flexible increase or decrease of CWMP instances, as the business volume changes, CWMP instances need to be increased or decreased. Adding or decreasing CWMP service instances in ACS will trigger an event, generate a new NGINX .conf file, replace the original .conf file, and finally restart the NGINX service.
7. According to claim 4, a method for implementing efficient scheduling and traffic diversion of an ACS system based on NGINX, characterized in that: The busy factor represents the busyness of the CWMP service instance, and is determined by the current number of connections of the CMWP service instance and the weight of the threshold.
8. According to the method of claim 1, which is characterized by: The ACS is an automatic configuration management system, which is responsible for managing the terminal equipment CPE in the TR protocol, isolating data of different service CPEs during management, and also manages the CWMP service, and is a window for operators; The CWMP is a component that interacts with the CPE device in the ACS; the CPE is a user-end device, or a managed device; the DISPATCH is a scheduling module, a custom NGINX module written based on LUA; the DNS is a domain name server that helps resolve URL parameters when the ACS and CPE use URL addresses to identify and access each other; the DHCP is a dynamic host configuration protocol server that assigns IP addresses to the ACS and CPE.
9. The TR protocol according to any one of claims 1 to 7, characterized in that: TR protocol is an application layer protocol located on top of IP. TR protocol defines the standard of southbound interface. The TR protocol data model standard includes the following: TR-069: Data Model Template for TR-069-Enabled Devices; TR-098: Internet Gateway Device Data Model for TR-069; TR-104: Provisioning Parameters for VoIP CPE; TR-181: Version 2 of the Device Data Model for TR-069.
10. A distributed ACS system, characterized in that: The distributed ACS system implements the efficient scheduling method based on NGINX as described in any one of claims 1-7 and satisfies the TR protocol.
