A 5G core network load balancing optimization method

Abstract

The present invention discloses a method for optimizing the load balancing of a 5G core network, including step 1: adding a new PCF2 node; step 2, the business processing process when PCF2 goes online; step 3, deleting the PCF2 node, that is, the business processing process when PCF2 goes offline; step 4, when PCF2 needs to be expanded, adding business components; step 5, when PCF2 needs to be scaled down, reducing business components. The method of the present invention optimizes the CPU utilization rate and avoids nodes with excessive loads; optimizes the space occupancy rate of storage resources, balances the IOPS of each node, optimizes the energy consumption monitoring, adjusts the load through migration, and reduces energy waste; the method of the present invention expands the registration and discovery mechanism, making it possible to have a performance distribution strategy based on specific parameters.

Description

Technical Field

[0001] The present invention belongs to the technical field of 5GC communication, and particularly relates to a method for optimizing the load balancing of a 5G core network. Background Art

[0002] Load Balancing (LB) is built on the existing network structure, providing a cheap, effective and transparent method to expand the bandwidth of network devices and servers, increase throughput, enhance network data processing capabilities, and improve network flexibility and availability. The traditional LB method is a basic implementation scheme of load balancing strategies, commonly used for load balancing of internal components of core network elements. Compared with the 4G core network, the 3GPP 5GC specification adopts a service-based interface and a registration and discovery mechanism to provide more perfect load balancing strategies for multiple nodes of network elements. A typical application is the NRF (network element providing registration and discovery functions) method. The load balancing scheme appears as an essential part of the network architecture. With the advancement of new technologies and new networks in 5G network construction and automated operation and maintenance, there is an objective requirement not to be limited to a single indicator such as load balancing of multiple network element nodes within a cluster, but to enable multi-dimensional dynamic control, such as dynamically adjusting the distribution of network elements participating in network processing among new and old nodes and new and old equipment manufacturers according to the operating indicators of network elements.

[0003] Patent application publication number CN108650685B discloses a method for optimizing the control plane of a C / U-separated 5G cellular heterogeneous network. In the C / U-separated architecture, macro base stations are responsible for control plane transmission, handling network control and wide coverage; small base stations are responsible for user plane transmission, offloading data for users at higher frequencies. After a macro base station fails, the coverage range of adjacent macro base stations of the failed macro base station is expanded according to requirements. Each small base station under the coverage area of the failed macro base station selects an adjacent macro base station to access and perform control plane transmission. The method for selecting adjacent macro base stations for small base stations is optimized. Based on the load balancing index and the maximum expandable coverage range of adjacent macro base stations, a statistical analysis and optimized deployment of macro base station selection for minimizing core network signaling load are formed.

[0004] With the development of cloud computing technology, load balancing increasingly represents a service that enhances the service capabilities of application systems for customers through traffic distribution, including improving system availability and avoiding single points of failure. Initially, load balancing provided network-layer balancing scenarios, achieving four-layer distribution based on TCP / UDP; later, the distribution capabilities for the application layer were added, that is, seven-layer traffic distribution capabilities, supporting application-layer protocols such as HTTP / HTTPS, and having elastic and large-scale distribution performance, realizing endogenous service integration capabilities, and further providing professional load balancing services. However, seven-layer load balancing is restricted by the protocols it supports, and checking HTTP headers consumes a large amount of system resources, which will inevitably affect system performance. In the case of a large number of connection requests, the load balancing device itself is likely to become a bottleneck in the overall network performance. Moreover, the operator's CT network is much more complex than IT: compared with 2G / 3G / 4G, the 5G network not only has more complex service functions, but also the introduction of service-oriented interfaces based on HTTP2 has changed the traditional signaling message communication architecture between network elements. To achieve message distribution at the service level in 2G / 3G / 4G networks, routing network elements are used, which is further conducive to interoperability between devices from different manufacturers and meets the requirements of network openness.

[0005] When the 2G / 3G / 4G network and the 5G network are mixedly networked, the network elements with message interaction architectures are divided into the following categories:

[0006] (1) Provider_NF: Network element that provides services, where NF is the network element;

[0007] (2) Consumer_NF: Network element that consumes services;

[0008] (3) Adjacent_NF: Related network elements in the interaction process between Provider_NF and Consumer_NF, such as service providers or callers of network management and databases, or third-party services;

[0009] (4) Routing_NF: That is, routing network elements, including DRA (Dynamic Resource Allocation) or 5G's SCP (Service Communication Proxy), or dedicated load balancing devices of external LBs. Of course, the routing_NF can be not configured, and the Consumer_NF selects the Provider_NF according to its own load balancing strategy.

[0010] In a communication network, there are the following three situations for load balancing requirements:

[0011] (1) The load balancing problem of new and old Provider_NFs in the 2G / 3G / 4G network, including the transition period of capacity expansion and adding new service nodes or replacing the equipment of old suppliers with new suppliers;

[0012] (2) The load balancing problem of the new and old Provider_NF in the 5G network, including the transition period of capacity expansion and adding new service nodes or replacing the equipment of the old provider with that of the new provider;

[0013] (3) The 2G / 3G / 4G network has the old Provider_NF, and the newly built 5G network has the new Provider_NF, including the transition period of capacity expansion and adding new service nodes or replacing the equipment of the old provider with that of the new provider.

[0014] At the same time, in all the above three situations, there is a cooperation problem between the new and old "Routing_NF". As an optional item, in order to uniformly schedule the new and old Provider_NF, the new Routing_NF will also connect to the old Provider_NF. In addition, of course, a new Routing_NF can also not be introduced for two reasons: 1) The Routing_NF requires more functions and even depends on changing the product architecture to provide. For example, the signaling forwarding function is defined as a new SCP network element in the 5G network and cannot be upgraded and supported on the 4G DRA network element; 2) When introducing the Provider_NF of the new provider, a new Routing_NF will be brought in accordingly to avoid the docking risk with the old Routing_NF.

[0015] Therefore, in the case of a 2G / 3G / 4G and 5G hybrid network, load balancing is very complex, unlike the IT network where only providing an external load balancing device can meet the load balancing requirements of application layer devices.

[0016] The advantage of NRF load balancing over LB is that even if the NRF becomes a failure point, it will not cause the communication interruption between NF consumers and NF providers. However, the disadvantage is that NF consumers and NF providers are fully topologically connected, making the routing configuration of each NF extremely complex. At the same time, the 4G and 5G converged network is the only way in the initial stage of 5G construction. Therefore, for a large-scale 4G / 5G network with a large number of NF nodes, a signaling Hub node needs to be established to simplify the signaling routing configuration and maintain the flexible control of the signaling routing.

[0017] The above two load balancing schemes do not consider the division of labor of NF (network elements) in a specific way. However, the wireless core network products and networking requirements including 5GC precisely include the division of labor of NF in a specific way. The disadvantage of the extended method of the load balancing function is that in some specific requirements, such as when 5G users are processed by new nodes and other users are processed by old nodes, LB and NRF still need to add the judgment of user categories. Load control NFs such as LB and NRF will be difficult to complete the work related to such services, and even some NF consumers will also be difficult to make judgments. Summary of the Invention

[0018] The object of the present invention is to provide a 5G core network load balancing optimization method capable of overcoming the above technical problems. The method of the present invention includes the following steps:

[0019] Step 1: Add a new PCF (Policy Control Function) 2 node:

[0020] Step 1.1: Build the software and hardware resources required for PCF2;

[0021] Step 1.2: Create a management agent. The management agent manages each PCF network element. On the one hand, the management agent is responsible for sending down control policy information, and the control policy information includes: NF provider list information, load balancing policy information, new service capability online information, service capability power-off information, and service processing capability adjustment information. On the other hand, the management agent completes the management message interaction with the LB and PCF;

[0022] Step 1.3: Add a management interface connected to the management agent on PCF1. The management interface is responsible for receiving the above control information sent down by the management agent and is responsible for providing the resources for real-time reporting of capacity and the performance statistics of the load on the management interface;

[0023] Step 1.4: PCF2 connects to the management agent and initiates a registration process to the LB, and then PCF2 waits to go online;

[0024] Step 1.5: Create a connection from the LB to the management agent on the LB, prepare to receive the control policy information from the management agent described in Step 1.2, and support the reporting of relevant performance statistics;

[0025] Step 2: Business processing process when PCF2 goes online:

[0026] Step 2.1: PCF2 notifies the management agent to process a service request, and the request message carries capacity information and processing capability information;

[0027] Step 2.2: The management agent synthesizes the capacity and processing capability information of PCF2 and PCF1 and, according to the load balancing policy sent down by the network management center, that is, all user accesses after a certain time period are processed by PCF2, notifies PCF2 to prepare to process the service, and notifies the LB how to distribute the service. The network management center is used to send down the load balancing policy, and the processing capability information includes: access and mobility-related policies, UE policies, and charging control policies;

[0028] Step 2.3: The LB distributes the PDU session policy control request service to PCF2. PCF2 starts to process the service and reports the real-time statistics of capacity and load to the management agent. The management agent determines that PCF2 is processing the service;

[0029] Step 3: Delete the PCF2 node, that is, the business processing process when PCF2 goes offline:

[0030] Step 3.1: PCF2 notifies the control agent that it is no longer responsible for the service processing network element, and reports to the control agent the relevant capacity and processing capacity information generated during the previous service processing period;

[0031] Step 3.2: The control agent synthesizes the capacity and processing capacity information of PCF2 and PCF1, and adjusts the service request distribution list according to the load balancing policy issued by the network management center, that is, the user access after a certain period is processed by PCF2, and it is changed to notify PCF1 to prepare for service processing and notify LB how to distribute services;

[0032] Step 3.3: LB cancels distributing services to PCF2 and changes to distributing services to PCF1 for processing, and reports the real-time statistical information of capacity and load to the control agent, and the control agent determines that PCF2 exits the service;

[0033] Step 4: When PCF2 requests capacity expansion and needs to add service components (a set of functional modules that support a certain service function), the following steps are included:

[0034] Step 4.1: Create new service components to undertake more service requests;

[0035] Step 4.2: The control component of PCF2 notifies the control agent that the new service components are ready for online, including transmitting the information of the increased processing capacity;

[0036] Step 4.3: When the control agent determines whether to enable the newly added capacity according to the latest load balancing policy, it notifies LB and PCF2. When the newly added capacity needs to be enabled, a soft handover method is adopted, that is, first enable the newly added capacity of the new node PCF2, and then notify the old node PCF1 to unload its relevant service components after the performance statistics of load balancing meet the requirements and remain stable, so as to reduce the processing function of PCF1;

[0037] Step 4.4: LB implements load balancing control according to the load adjustment information transmitted by the control agent and PCF2, and distributes more service requests to PCF2;

[0038] Step 4.5: After the new service components of PCF2 meet the stable operation requirements, the new capacity is confirmed between PCF2 and the control agent, and the entire service processing system including PCF1, PCF2 and the control agent reaches a steady state;

[0039] Step 4.6: The control agent decides whether to make an effective capacity adjustment to other existing resources of PCF1 including memory and hard disk resources according to the latest load balancing policy. When adjustment is needed, it is achieved by unloading the relevant service components on PCF1 again;

[0040] Step 5: When PCF2 needs to downsize and reduce service components, the following steps are included:

[0041] Step 5.1: Determine the resources of the service components to be taken offline;

[0042] Step 5.2: The control component of PCF2 notifies the management agent that the components of the service to be taken offline are ready for taking offline, and transmits to the management agent information on how much processing capacity has been reduced;

[0043] Step 5.3: The management agent determines whether to accept capacity reduction according to the latest load balancing policy and notifies LB and PCF2. When accepting capacity reduction, the soft handover method is adopted, that is, first enable the alternative capacity of PCF1, and after the performance statistics of load balancing meet the requirements and remain stable, then notify PCF2 to unload its relevant service components;

[0044] Step 5.4: LB implements load balancing control according to the load adjustment information transmitted by the management agent and PCF2, and correspondingly reduces the service requests distributed to PCF2;

[0045] Step 5.5: After the service components of PCF2 meet the stable operation requirements, the new capacity is confirmed between PCF2 and the management agent, and the entire service processing system reaches a steady state;

[0046] Step 5.6: PCF2 releases the resources occupied by the corresponding part of the service components.

[0047] The method of the present invention has the following beneficial effects:

[0048] 1. The method of the present invention has specialized division of labor, and the load balancing function is completed by an independent NF or an independent component;

[0049] 2. The method of the present invention has little negative impact on availability, maximally improves the availability of the NF cluster. On the one hand, it avoids single point of failure, and on the other hand, it can maximize the distribution performance;

[0050] 3. The method of the present invention can distribute flexibly. From the perspective of resources, load balancing is called DRS, that is, dynamic resource balancing, which optimizes CPU utilization and avoids nodes with excessive load; it optimizes the space occupancy rate of storage resources, balances the IOPS of each node, optimizes energy consumption monitoring, adjusts the load through migration, and reduces energy waste;

[0051] 4. The method of the present invention can distribute accurately. In some application scenarios, it is necessary to accurately distribute to specific nodes. For example, the response message must return to the processing node of the request message for processing. Another example is the situation where there is a corresponding relationship between the node and the user or session. When selecting a specific user or session, it must be distributed to the node, and other nodes cannot process it;

[0052] 5. The method of the present invention extends the registration discovery mechanism, making it possible to have a performance distribution strategy based on specific parameters. Detailed implementation manner

[0053] The following describes the implementation manner of the present invention in detail. The method of the present invention includes the following steps:

[0054] Step 1: Add a new PCF (Policy Control Function) 2 node:

[0055] Step 1.1: Build the software and hardware resources required for PCF2;

[0056] Step 1.2: Create a management agent. The management agent manages each PCF network element. On the one hand, the management agent is responsible for sending down control policy information, and the control policy information includes: NF provider list information, load balancing policy information, new service capability online information, service capability power-off information, and service processing capability adjustment information. On the other hand, the management agent completes the management message interaction with the LB and PCF;

[0057] Step 1.3: Add a management interface connected to the management agent on PCF1. The management interface is responsible for receiving the above control information sent down by the management agent and is responsible for providing resources for real-time reporting of capacity and performance statistics of the load on the management interface;

[0058] Step 1.4: PCF2 connects to the management agent and initiates a registration process to the LB, and then PCF2 waits to go online;

[0059] Step 1.5: Create a connection from the LB to the management agent on the LB, prepare to receive the control policy information from the management agent described in Step 1.2, and support the reporting of relevant performance statistics;

[0060] Step 2: Business processing process when PCF2 goes online:

[0061] Step 2.1: PCF2 notifies the management agent to process a service request, and the request message carries capacity information and processing capability information;

[0062] Step 2.2: The management agent synthesizes the capacity and processing capability information of PCF2 and PCF1 and, according to the load balancing policy sent down by the network management center, that is, all user accesses after a certain period of time are processed by PCF2, notifies PCF2 to prepare to process the service, and notifies the LB how to distribute the service. The network management center is used to send down the load balancing policy, and the processing capability information includes: access and mobility-related policies, UE policies, and charging control policies;

[0063] Step 2.3: LB distributes the PDU session policy control request service to PCF2. PCF2 starts to process the service and reports the real-time statistics of capacity and load to the management agent. The management agent determines that PCF2 is processing the service;

[0064] Step 3: The process of deleting the PCF2 node, that is, the business processing when PCF2 goes offline:

[0065] Step 3.1: PCF2 notifies the management agent that it is no longer responsible for the service processing network element and reports the relevant capacity and processing capacity information generated during the previous service processing to the management agent;

[0066] Step 3.2: The management agent synthesizes the capacity and processing capacity information of PCF2 and PCF1, and adjusts the service request distribution list according to the load balancing policy issued by the network management center. That is, the user access after a certain period of time is changed from being processed by PCF2 to notifying PCF1 to prepare to process the service and notifying LB how to distribute the service;

[0067] Step 3.3: LB cancels the distribution of services to PCF2 and changes to distribute the services to PCF1 for processing, and reports the real-time statistics of capacity and load to the management agent. The management agent determines that PCF2 exits the service;

[0068] Step 4: When PCF2 needs to expand its capacity and add service components (a set of functional modules that support a certain service function), the following steps are included:

[0069] Step 4.1: Create new service components to undertake more service requests;

[0070] Step 4.2: The control component of PCF2 notifies the management agent that the new service components are ready for online, including transmitting the information of the increased processing capacity;

[0071] Step 4.3: The management agent judges whether it is necessary to enable the newly added capacity according to the latest load balancing policy, and notifies LB and PCF2. When it is necessary to enable the newly added capacity, the soft handover method is adopted, that is, first enable the newly added capacity of the new node PCF2, and then notify the old node PCF1 to unload its relevant service components after the performance statistics of load balancing meet the requirements and remain stable, so as to reduce the processing function of PCF1;

[0072] Step 4.4: LB implements load balancing control according to the load adjustment information transmitted by the management agent and PCF2, and distributes more service requests to PCF2;

[0073] Step 4.5: After the new service components of PCF2 meet the requirements for stable operation, the new capacity is confirmed between PCF2 and the management agent, and the entire service processing system including PCF1, PCF2, and the management agent reaches a steady state;

[0074] Step 4.6: The management agent decides whether to make an effective capacity adjustment to other existing resources of PCF1 that include memory and hard disk resources according to the latest load balancing policy. When adjustment is needed, it is achieved by uninstalling the relevant service components on PCF1 again;

[0075] Step 5: When PCF2 needs to downsize and reduce service components, the following steps are included:

[0076] Step 5.1: Determine the resources of the service components to be taken offline;

[0077] Step 5.2: The control component of PCF2 notifies the management agent that the service components to be taken offline are ready for offline and transmits to the management agent the information on how much processing capacity has been reduced;

[0078] Step 5.3: The management agent judges whether to accept the capacity reduction according to the latest load balancing policy and notifies LB and PCF2. When accepting the capacity reduction, it adopts the soft handover method, that is, first enables the alternative capacity of PCF1, and then notifies PCF2 to uninstall its relevant service components after the performance statistics of load balancing meet the requirements and remain stable;

[0079] Step 5.4: LB implements load balancing control according to the load adjustment information transmitted by the management agent and PCF2, and correspondingly reduces the service requests distributed to PCF2;

[0080] Step 5.5: After the service components of PCF2 meet the requirements for stable operation, the new capacity is confirmed between PCF2 and the management agent, and the entire service processing system reaches a steady state;

[0081] Step 5.6: PCF2 releases the resources occupied by the corresponding part of the service components.

[0082] Taking the automatic activation of VoIP services for user terminals in a mobile network as an example, for the mobile terminal APP to obtain VoIP or VoLTE services provided by the IMS (IP Multimedia Subsystem) network, it first needs to register in the IMS network and obtain relevant registration information, including the IP information of the P-CSCF (Proxy Call Session Control Function), username, and password. The service process is as follows:

[0083] Step 101, the APP establishes an HTTP service process with Prov_Ser (registration agent, providing registration agent services for the user data center), and sends an HTTP Get message, the message body of which contains the MSISDN and IMSI numbers;

[0084] Step 202, Prov_Ser generates an IMPU (User Public Identity) and an IMPI (User Private Identity), as well as a username and password based on the MSISDN (Mobile Subscriber Identification Number) and IMSI (International Mobile Subscriber Identity);

[0085] Step 303, Prov_Ser generates an MML (Man-Machine Language) instruction and sends a request for modifying the authentication information of the PVI (Private Identity) and updates the username and password to the UDC (User Data Center);

[0086] Step 404, the UDC returns a modification success message to Prov_Ser;

[0087] Step 505, Prov_Ser returns an HTTP response message to the terminal APP, including the username and password, as well as P-CSCF IP information.

[0088] After the above process from Step 101 to Step 505 is completed, the mobile terminal APP uses the username and password to initiate a registration with the P-CSCF to complete the authentication process for registering to the IMS network. When introducing a new UDC2, it is required that the MML (Man-Machine Language) instruction realizes load balancing between the new and old UDC nodes. The user scopes served by the new UDC2 and the old UDC1 are different, and different IMSI number segments are allocated. As the capacity of the new UDC increases, there are requirements for dynamic and manual intervention in the load balancing strategy.

[0089] Adopt a load balancing solution that combines LB and proxy Agent settings. LB and Agent shield the impact of multiple UDCs, that is, multiple service providers, from Prov_Ser. Compared with a single-node UDC, the difference in the numbering service flow when Prov_Ser initiates a request for modifying the authentication information of a certain PVI is that LB and Agent realize the intelligent distribution of MML instructions.

[0090] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the scope disclosed by the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. A method for optimizing the load balancing of a 5G core network, characterized in that, It includes the following steps: Step 1: Add a new PCF2 node: Step 1.1: Build the software and hardware resources required for PCF2; Step 1.2: Create a control agent. The control agent manages each PCF network element. On the one hand, the control agent is responsible for distributing control policy information, which includes: NF provider list information, load balancing policy information, new service capability online information, service capability power-off information, and service processing capability adjustment information. On the other hand, the control agent completes the management message interaction with the LB and PCF; Step 1.3: Add a management interface connected to the control agent on PCF1. The management interface is responsible for receiving the control policy information distributed by the control agent and providing performance statistics of resources and loads for real-time reporting on the management interface; Step 1.4: PCF2 connects to the control agent and initiates a registration process to the LB, and then PCF2 waits to go online; Step 1.5: Create a connection from the LB to the control agent on the LB, prepare to receive the control policy information from the control agent in Step 1.2, and support related performance statistics reporting; Step 2: Business processing process when PCF2 goes online; Step 2.1: PCF2 notifies the control agent to process a service request, and the request message carries capacity information and processing capability information; Step 2.2: The control agent synthesizes the capacity and processing capability information of PCF2 and PCF1 and, according to the load balancing policy issued by the network management center, that is, all user accesses after a certain period of time are processed by PCF2, notifies PCF2 to prepare to process the service, notifies the LB how to distribute the service. The network management center is used to issue the load balancing policy, and the processing capability information includes access and mobility-related policies, UE policies, and charging control policies; Step 2.3: The LB distributes the PDU session policy control request service to PCF2. PCF2 starts to process the service and reports real-time capacity and load statistics information to the control agent. The control agent determines that PCF2 is processing the service; Step 3: Delete the PCF2 node, that is, the business processing process when PCF2 goes offline; Step 3.1: PCF2 notifies the control agent that it is no longer responsible for the service processing network element and reports the relevant capacity and processing capability information generated during the previous service processing to the control agent; Step 3.2: The control agent synthesizes the capacity and processing capability information of PCF2 and PCF1 and adjusts the service request distribution list according to the load balancing policy issued by the network management center, that is, the user access after a certain period of time is processed by PCF2, and changes to notify PCF1 to prepare to process the service and notify the LB how to distribute the service; Step 3.3: The LB cancels distributing the service to PCF2 and changes to distribute the service to PCF1 for processing, and reports real-time capacity and load statistics information to the control agent. The control agent determines that PCF2 exits the service; Step 4: When PCF2 needs to expand its capacity, business components need to be added; Step 4.1: Create new business components to undertake more service requests; Step 4.2: The control component of PCF2 notifies the control agent that the new business components are ready for going online to transfer the increased processing capability information; Step 4.3: When the control agent determines whether to enable the newly added capacity according to the latest load balancing policy and notifies LB and PCF2, if it is necessary to enable the newly added capacity, a soft handover method is adopted, that is, the newly added capacity of the new node PCF2 is first enabled, and after the performance statistics of the load balancing meet the requirements and remain stable, the old node PCF1 is notified to unload its related service components to reduce the processing function of PCF1; Step 4.4: LB implements load balancing control according to the load adjustment information transmitted by the control agent and PCF2, and distributes more service requests to PCF2; Step 4.5: After the new service components of PCF2 meet the stable operation requirements, the new capacity is confirmed between PCF2 and the control agent, and the entire service processing system including PCF1, PCF2, and the control agent reaches a steady state; Step 4.6: The control agent decides whether to effectively adjust the existing resources including memory and hard disk resources of PCF1 according to the latest load balancing policy. If adjustment is required, it is achieved by unloading the related service components on PCF1; Step 5: When PCF2 needs to reduce its capacity and reduce service components; Step 5.1: Determine the resources of the service components to be taken offline; Step 5.2: The control component of PCF2 notifies the control agent that the service components to be taken offline are ready for taking offline and transmits the information on how much processing capacity has been reduced to the control agent; Step 5.3: When the control agent determines whether to accept the capacity reduction according to the latest load balancing policy and notifies LB and PCF2, if it accepts the capacity reduction, a soft handover method is adopted, that is, the alternative capacity of PCF1 is first enabled, and after the performance statistics of the load balancing meet the requirements and remain stable, PCF2 is notified to unload its related service components; Step 5.4: LB implements load balancing control according to the load adjustment information transmitted by the control agent and PCF2, and correspondingly reduces the service requests distributed to PCF2; Step 5.5: After the service components of PCF2 meet the stable operation requirements, the new capacity is confirmed between PCF2 and the control agent, and the entire service processing system reaches a steady state; Step 5.6: PCF2 releases the resources occupied by the corresponding part of the service components.