Network element resource allocation method and system for power IMS (IP Multimedia Subsystem) virtualized core network

Abstract

The invention discloses a network element resource allocation method and system for a power IMS (IP Multimedia Subsystem) virtualization core network, and belongs to the technical field of power IMS network function virtualization. The method comprises the following steps: S1, establishing a hybrid networking model of a core network, and determining simplest node configuration and corresponding network element nodes of the core network based on the model; s2, dividing the network element nodes into key network elements and non-key network elements, and configuring single node redundancy for the network element nodes based on the division and a preset redundancy coefficient; and S3, configuring a synchronizable instance for the network element node configured with the single-node redundancy so as to form double-node redundancy configuration of the core network. According to the method, a resource configuration system from simplest configuration to multiple redundancy is constructed step by step, so that efficient flow management and user cutover can be realized, and the service persistence and the overall robustness of the system are effectively optimized.

Description

Technical Field

[0001] This invention relates to the field of communication network technology, and in particular to a method for configuring network element resources in a power IMS (IP Multimedia Subsystem) virtualized core network, specifically applicable to power IMS core network scenarios employing Network Functions Virtualization (NFV) technology. Background Technology

[0002] With the development of power communication networks, administrative switching and dispatch communication services based on IMS have placed higher demands on network reliability, flexibility, and resource utilization efficiency. Network Function Virtualization (NFV) technology enables flexible deployment and elastic scaling of the power IMS core network by software-firing the functions of traditional dedicated hardware network elements (PNFs) into virtual network functions (VNFs) and deploying them on a general-purpose server resource pool.

[0003] However, in the practice of NFV core network for power IMS, configuring network element resources efficiently and reliably faces a series of challenges. First, the core network contains various functional entities (such as CSCF and HSS), whose virtualized instances (VNFs) need to be deployed on physical servers. Existing methods often lack a systematic hybrid networking model to coordinate the topological relationships between physical and virtual network elements, as well as switching and routing devices. This makes it difficult to quickly determine a minimal node configuration while meeting service traffic and resource constraints, potentially leading to redundant or insufficient initial resource configurations, impacting deployment efficiency and cost.

[0004] Secondly, to ensure high reliability of power production communications, the core network must possess a certain degree of fault tolerance. Current technologies typically employ a rather crude strategy when configuring redundancy for network elements, failing to differentiate between critical and non-critical elements based on their importance to service continuity, and thus failing to impose differentiated redundancy requirements. This can lead to insufficient redundancy for critical elements and excessive redundancy for non-critical elements, making it difficult to achieve an optimal balance between system robustness and resource costs, i.e., hindering the implementation of optimized single-node redundancy configurations.

[0005] Furthermore, when network load changes dynamically or a single point of failure occurs, rapid recovery or expansion of service capabilities is required. Existing solutions have shortcomings in achieving seamless service switching and dynamic scaling. On the one hand, the lack of an efficient multi-active instance synchronization mechanism can lead to potentially long service interruption times during failover; on the other hand, dynamic scaling strategies are often rigid and fail to effectively combine load awareness with service cutover, making it difficult to achieve smooth and elastic resource scaling while ensuring service continuity, i.e., it is difficult to efficiently build and maintain dual-node redundant configurations.

[0006] In summary, existing technologies lack a systematic approach that can begin by constructing a basic network model, progressively optimize resource allocation, and ultimately form a power IMS virtualization core network with high reliability, high resource utilization, and rapid resilience. Therefore, there is an urgent need to propose a new method for configuring network element resources to address these issues. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method for configuring network element resources of power IMS virtualization core network. This method can systematically perform the entire process of resource configuration from basic modeling to high reliability architecture construction, realize efficient traffic management and user cutover, and significantly optimize service continuity capability and overall system robustness.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0009] A method for configuring network element resources in a power IMS virtualization core network includes the following steps:

[0010] S1. Establish a hybrid networking model for the core network, and determine the simplest node configuration and corresponding network element nodes for the core network based on the model;

[0011] S2. Divide the network element nodes into critical network elements and non-critical network elements, and configure single-node redundancy for the network element nodes based on the division and a preset redundancy coefficient;

[0012] S3. Configure a synchronous instance for a network element node that has been configured with single-node redundancy to form a dual-node redundancy configuration for the core network.

[0013] In one embodiment, step S1 includes: constructing an optimization problem based on the hybrid networking model with the objective of minimizing the number of network element nodes; solving the optimization problem using a greedy algorithm to obtain the simplest node configuration.

[0014] In one embodiment, the hybrid networking model includes a set of nodes consisting of physical network elements, virtual network elements, and connecting devices, as well as a set of edges representing connections between nodes; the constraints of the optimization problem include virtual machine resource constraints and traffic conservation constraints.

[0015] Further, in step S1, the hybrid networking model includes: a node set V, including PNFs, VNFs, Layer 2 switches, Layer 3 routers, CE devices, and a network management system; a virtual network element instance v, which is an instance of the VNF and mapped to a virtual machine; and an edge set E, including virtual links between the VNFs and physical links between the PNFs. The optimization objective of the hybrid networking model is established as follows: under the conditions of satisfying virtual machine resource constraints and traffic conservation constraints, minimize the number of network element nodes contained in the node set V in the core network topology graph G=(V,E); the virtual machine resource constraints are: for the virtual network element instance v in the network element node, the CPU and memory resource requirements of its corresponding VNF do not exceed the CPU and memory resource capacity allocated by the server to the virtual machine; and the traffic conservation constraints are: the total inflow through the network element node is equal to the sum of the total outflow of the network element node and the storage change of the server.

[0016] Furthermore, in step S1, the simplest node configuration of the core network is solved by a greedy algorithm, specifically as follows: obtain the resource requirement function values ​​of all network elements and generate a network element requirement set; obtain the CPU and memory resources of available servers and generate a resource pool; input the network element requirement set and the resource pool; obtain the server ID assigned to each network element by a greedy algorithm, generate a resource configuration list and output it.

[0017] In one embodiment, step S2 includes: determining the basic resource configuration based on the simplest node configuration; calculating the network element redundancy increment based on the redundancy coefficient of the key network element and the redundancy coefficient of the non-key network element; and solving the single-node redundancy configuration by a recursive algorithm, with the goal of minimizing the deployment cost of computing resources, in combination with the basic resource configuration and the network element redundancy increment.

[0018] Furthermore, in step S2, the optimization objective of the core network single-node redundancy configuration is set as follows: minimizing the computing resource deployment cost under the condition that the selected network element nodes do not exceed the upper limit of VNF resource requirements and the system redundancy meets the safe operating rate.

[0019] Furthermore, in step S2, the core network single-node redundancy configuration is solved using a recursive algorithm, specifically as follows: the basic configuration is obtained based on the simplest node configuration of the core network; redundancy coefficients are set for the critical network elements and the non-critical network elements respectively; the network element redundancy increment for each resource of the core network is calculated, and an initial step size and availability threshold are set; the obtained set of network element nodes, the basic configuration, the initial step size, and the availability threshold are input; the redundancy level and the corresponding single-node redundancy configuration are obtained and output using a recursive algorithm.

[0020] In one embodiment, step S3 includes: establishing a multi-active instance synchronization mechanism to switch the services of a network element node to a synchronizeable instance of another network element node when a failure occurs.

[0021] In one embodiment, when the multi-active instance synchronization mechanism is triggered, the following operations are performed: a target synchronization instance is selected from the remaining network element nodes; single-node redundancy is configured for the target synchronization instance; and the services of the faulty network element node are switched to the target synchronization instance.

[0022] In one embodiment, step S3 further includes: implementing load-aware dynamic capacity expansion, and selecting to perform VNF capacity expansion or service cutover operation based on the capacity ratio of Virtual Network Functions (VNFs).

[0023] In one embodiment, when the VNF capacity percentage is lower than a first threshold, a VNF expansion operation is performed; when the VNF capacity percentage is not lower than the first threshold, a service cutover operation is performed.

[0024] In one embodiment, the dynamic expansion process also includes an automatic rollback mechanism, which automatically terminates the current operation and switches the service back to its original state when the expansion or cutover fails.

[0025] Furthermore, in step S3, the multi-active instance synchronization method specifically involves: establishing a synchronization mechanism; when a network element node in the single-node redundant configuration fails and there are synchronizable instances among the remaining network elements, triggering the synchronization mechanism to select a target synchronization instance; executing the recursive algorithm in step S3 on the target synchronization instance to obtain its single-node redundant configuration; and switching the failed network element node to the target synchronization instance.

[0026] Furthermore, in step S3, the dynamic expansion algorithm specifically comprises: when the VNF capacity ratio is <60%, performing VNF expansion, creating a new VNF instance, migrating users, and verifying the migration result; when the VNF capacity ratio is ≥60%, performing service cutover, adjusting the PNF-VNF service boundary, and performing service cutover within a preset maintenance window and verifying the cutover result; if expansion fails due to excessive service interruption time, an automatic rollback mechanism is triggered: terminating the newly created VNF instance; switching the migrated users back to the original network element node; sending an alarm notification to the operation and maintenance management terminal; and recording the corresponding fault code in the knowledge base.

[0027] The present invention also provides a network element resource configuration system for a power IMS virtualization core network, including a processing unit configured to execute the network element resource configuration method as described in any of the preceding claims.

[0028] Compared with the prior art, the present invention has the following significant advantages:

[0029] Systematic modeling and initial configuration optimization: By establishing a hybrid network model that includes physical and virtual elements and using a greedy algorithm to solve for the simplest node configuration, a resource-efficient and structurally clear basic deployment blueprint for the network is provided, reducing initial investment costs and operational complexity.

[0030] Differentiated redundancy and cost optimization: By intelligently dividing critical and non-critical network elements and assigning them different redundancy coefficients, a recursive algorithm is used to solve for the optimal redundancy level under the constraint of safe operating rate. This method achieves the best balance between reliability goals and resource costs, avoiding "overprotection" or "underprotection".

[0031] High-level business continuity and elasticity assurance: By integrating a multi-active instance synchronization mechanism and a load-aware dynamic scaling algorithm, it can not only achieve fast and smooth business switching (forming dual-node redundancy) when a node fails, but also intelligently trigger scaling or cutover according to business load, and has the ability to automatically roll back in case of failure, which greatly improves the system's service continuity, adaptability and overall robustness. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of the core network topology in one embodiment of the present invention.

[0033] Figure 2 This is a flowchart of solving the simplest node configuration using a greedy algorithm in one embodiment of the present invention.

[0034] Figure 3 This is a schematic diagram illustrating the implementation of a differentiated management strategy for network element nodes in one embodiment of the present invention.

[0035] Figure 4 This is a flowchart of a recursive algorithm for solving a single-node redundancy configuration in one embodiment of the present invention.

[0036] Figure 5 This is a schematic block diagram of a multi-active instance synchronization mechanism in one embodiment of the present invention.

[0037] Figure 6 This is a schematic block diagram of a load-aware dynamic expansion algorithm in one embodiment of the present invention.

[0038] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. It should be noted that the following embodiments are only for explaining the present invention and are not intended to limit the present invention.

[0040] The network element resource configuration method for the power IMS virtualized core network provided by this invention adopts a three-step progressive strategy: first, laying the foundation (obtaining the simplest feasible configuration through models and algorithms); then, strengthening (differentiating redundancy based on the importance of network elements); and finally, enhancing the system (introducing multi-active synchronization and dynamic elasticity mechanisms to achieve high-level reliability). Figure 1 The diagram shows a typical core network topology to which this method can be applied, which includes a PNF pool, a VNF pool, Layer 2 / Layer 3 switching and routing devices, and CE devices, together forming a hybrid network environment.

[0041] Step S1: Establish a hybrid network model and determine the simplest node configuration.

[0042] This step aims to provide a quantitative optimization basis for network planning. First, a hybrid network model G = (V, E) for the core network is established. Here, the node set V includes all physical network elements (PNFs), virtual network elements (VNFs), Layer 2 switches (L2Switch), Layer 3 routers (L3Router), CE devices, and the network management system (OMC). Each VNF instance v maps to a specific virtual machine (VM). The edge set E includes virtual links between VNF instances and physical links between PNFs.

[0043] Based on this model, an optimization problem is constructed. The optimization objective is to minimize the total number of network element nodes |V'| required to be activated in the network topology, while satisfying the following two core constraints.

[0044] Virtual machine resource constraints: For any VNF instance v, the CPU and memory resource requirements of its VNF must not exceed the maximum CPU and memory capacity allocated to the VM by the host server.

[0045] Traffic conservation constraint: For any network element node (especially a forwarding device), its total inbound traffic is equal to the sum of the total outbound traffic and the change in storage within the node (such as a server).

[0046] To solve this optimization problem, this invention employs a greedy algorithm. For example... Figure 2 As shown, the specific process is as follows:

[0047] Input: The set of resource requirement function values ​​D for all network elements to be deployed, and the available server resource pool S (including the remaining CPU and memory information of each server).

[0048] Process: The algorithm iteratively selects the network element d with the "most urgently needed resource" (e.g., the largest weighted sum of CPU and memory requirements) from the demand set D, then finds the server s from the resource pool S that can meet its requirements and has the "best match" of remaining resources (e.g., the smallest resource fragmentation). The network element d is deployed to the server s, and the remaining resources of the server s are updated. This process is repeated until all network elements are deployed or resources are insufficient.

[0049] Output: A resource configuration list A, recording the server ID ultimately assigned to each network element.

[0050] The deployment scheme obtained through this algorithm is the "simplest node configuration" under the current resource constraints, which forms the basis for all subsequent redundancy and elasticity designs.

[0051] Step S2: Divide critical / non-critical network elements and configure single-node redundancy.

[0052] After obtaining the minimum configuration, redundancy needs to be added to improve reliability. This invention does not add redundancy equally to all network elements, but rather adopts a differentiated management strategy. For example... Figure 3 As shown, the network element set X in the simplest configuration is first divided into a key network element set. (such as core control plane network elements like CSCF and HSS) and non-critical network element sets (such as some MRF media surface elements).

[0053] Set a redundancy coefficient λ for each. n (Key network elements) and u n (Non-critical network elements), and λ n >u n >0. Based on this, calculate the network element redundancy increment Δ for the l-th resource (e.g., CPU). l (u), whose calculation formula takes into account the differentiated contributions of different types of network elements.

[0054] Subsequently, a global redundancy level t (an integer t ≥ 1) is defined. Then, the resource configuration Fr for a single node with redundancy is determined. l (u)=F l (u)+t·△ l (u),l=1,2,…,L(12), where Fr l (u) represents the resource requirements under the simplest configuration.

[0055] The optimization goal of this step is to meet the resource configuration limit (the demand after redundancy does not exceed the total physical resources) and the system's safe operating rate (such as availability not being lower than the set threshold P). *Under two constraints, minimize the redundancy level t, thereby minimizing the additional computing resource deployment cost caused by redundancy. This is a constrained integer optimization problem.

[0056] like Figure 4 As shown, this invention uses a recursive algorithm to solve this problem:

[0057] Input: Minimal configuration X * Basic resource requirements Fr l (u), Redundancy increment △ l (u), initial step size (e.g., t=1), availability threshold P * .

[0058] Process: The algorithm starts from t=1 and calculates the configuration Fr at the current t. l (u) and the estimated system safety operating rate if Then, increment t by one step (recursive call) and recalculate; if the constraint is satisfied, record this t value. Through recursive search, find the minimum t value that satisfies the safe operating rate constraint.

[0059] Output: The optimal redundancy level t* and its corresponding single-node redundancy configuration scheme Fr l (u). This configuration ensures that the system can still operate by relying on its local redundant resources when any network element node fails.

[0060] Step S3: Configure a synchronized instance to form a dual-node redundancy configuration.

[0061] To achieve a higher level of availability (such as tolerating the failure of multiple instances of the same service), a cross-node active-active or multi-active architecture needs to be built on top of single-node redundancy. This step is achieved through two parts: multi-active instance synchronization and dynamic scaling.

[0062] 1. Multi-active instance synchronization method:

[0063] like Figure 5As shown, a distributed synchronization mechanism is established. Session state information is synchronized and managed with distributed locks through distributed key-value stores such as ETCD; media streams support dual-path transmission of RTP / RTCP; configuration data is incrementally synchronized using rsync. This mechanism is triggered when the monitoring system detects a failure of a network element node x_n in the single-node redundancy configuration. The system first selects the most up-to-date and lightest-loaded synchronized instance from the surviving synchronizeable instances on other physical nodes as the target synchronization instance. Then, it immediately executes the recursive algorithm of step S2 on the target synchronization instance to quickly calculate and load a single-node redundancy configuration of the same level as the failed node. Finally, the service traffic of the failed node (such as through updating DNS records or load balancer configuration) is smoothly switched to the hardened target synchronization instance. Thus, the system recovers from the "single-node redundancy" state to the "dual-node redundancy" state.

[0064] 2. Load-aware dynamic scaling algorithm:

[0065] To ensure the system always has a sufficient resource pool to provide "synchronizable instances" and to cope with business growth, dynamic scaling is required. For example... Figure 6 As shown, the algorithm continuously monitors the overall capacity percentage of the VNF pool.

[0066] When the capacity utilization rate is less than 60% (first threshold): resources are considered relatively sufficient, and VNF expansion is performed. This involves creating a new VNF instance in the resource pool, migrating some user sessions to the new instance, and verifying whether the service is normal after the migration.

[0067] When the capacity utilization rate is ≥60%, it is considered that resources are becoming strained, and service cutover will be performed. This means that the service boundary between PNF and VNF needs to be adjusted, which may involve migrating some services from PNF to VNF, or horizontally scaling the VNF cluster. This operation is performed within a preset maintenance window, and rigorous verification is conducted after the cutover is completed.

[0068] Automatic rollback mechanism: Regardless of whether it is expansion or cutover, if it fails for any reason (such as service interruption timeout), an automatic rollback will be triggered: immediately terminate the newly created instance or operation; switch the migrated users back to the original stable node; send an alarm to the operation and maintenance platform; and record the code and context of this failure to the knowledge base for subsequent analysis and optimization.

[0069] Through step S3, this invention not only achieves a rapid upgrade from N+1 redundancy to a dual-active architecture in case of failure, but also ensures the effective supply of redundant resources and smooth business growth through intelligent dynamic resource scheduling, thereby forming a core network dual-node redundancy configuration with elasticity and self-healing capabilities.

[0070] Specifically:

[0071] The core network topology used in this invention is as follows: Figure 1 As shown. The core network uses Layer 2 switches (L2Switches) for east-west traffic aggregation. Communication between VNFs is isolated by VLANs on the Layer 2 switches. At the same time, Layer 3 routers (L3Routers) are used for north-south traffic forwarding, connecting to CE routers (CERouters). User access traffic is distributed to different network element pools through the PBR (Policy-Based Routing) of the Layer 3 routers. PNFs (Physical Network Functions) and VNFs are grouped into pools to handle services in different areas, and their OMCs are configured independently. During user cutover, new users are preferentially assigned to VNF pools, and old users are migrated in batches.

[0072] In the IMS administrative switching lightweight NFV core network, the nodes of the main functional entities such as CSCF (Call Session Control Function), BGCF (Breakout Gateway Control Function), MGCF (Media Gateway Control Function), and HSS (Home Subscriber Server) contain multiple network elements, denoted by the set X = {x1, x2, ..., x...}. n ,…,x N} represents the set of network elements in the system, where x n This represents the nth network element in the system, where N represents the total number of network elements.

[0073] NFV technology allows network resources to be dynamically adjusted according to demand. When the number of users increases, the core network can automatically expand to support more calls and avoid fault tolerance bottlenecks. Therefore, the number of core network users is used as a decision variable, represented by U = {u1, u2, ..., u...}. m ,…,u M} represents the sample set of core network users, where u m Let represent the number of users corresponding to the m-th sample, and M represent the total number of samples. This indicates that when the number of core network users is u m At that time, network element x n The set of requirements for system hardware and software resources (CPU, memory, network bandwidth, virtual machine parameters, etc.), where r l n (u m ) represents the core network element x n The demand for the l-th type of resource, where L represents the number of resource types.

[0074] When the number of network elements, users, and network element resource requirements in the system are known prior from testing and data collection, and the testing period is long enough to ensure that the data has a clear distribution characteristic, in order to establish the relationship between NFV network element requirements and the number of users, and further predict the network element resource requirements under a specific number of users, the network element resource requirement function can be approximated using the least squares method. First, the following K-degree polynomial is constructed:

[0075]

[0076] In the formula, f l n (u m ) represents the network element x when the number of users is . n For resource r l n (u m The required fit value is about the variable u. m A function of degree K, using Let K+1 be a (K+1)×1 dimensional vector of undetermined coefficients, where Let represent the (k+1)th undetermined coefficient in the vector, (·). T This indicates the transpose operation.

[0077] To find the vector of undetermined coefficients The optimal estimates of each coefficient within the equation result in the resource demand fit value f. l n (u m The closest r obtained from actual testing l n (u m The following objective can be established for the unconstrained linear least squares problem:

[0078]

[0079] In the formula, n, l, K, u m and r l n (u m ) is a known quantity. and f l n (u m ) represents the objective to be solved.

[0080] To obtain the network element resource demand function f l n (u m The best fit is derived first through linear regression. The analytical solution involves substituting the known M sample values ​​of network element resource requirements with K-dimensional features into formula (1) and establishing the following structure:

[0081]

[0082] In the formula, f l n Let (U) be a K×1 dimensional vector, where the elements represent the fitted values ​​of network element resource requirements, and let U be an M×(K+1) dimensional matrix. Then the matrix of the objective loss function is:

[0083]

[0084] Among them, V l n =[r l n (u1),r l n (u2),…,r l n (u M )] T This is the output vector of the sample, with dimension M×1. Based on the principle of least squares, the loss function... about Taking the derivative and setting it to 0, we get:

[0085]

[0086] When matrix U T When U is reversible, The analytical solution is the vector under the optimal estimate, which can be expressed as:

[0087]

[0088] Furthermore, the network element resource demand function f l n (u m The expression for the optimal fit can be written as:

[0089]

[0090] Therefore, network element x n The formula for calculating the demand function value of the l-th resource in the system when the number of concurrent users u>0 is:

[0091]

[0092] Where K represents the degree of the fitting function used. This formula can be used to predict the core network element resource requirements under any number of users within a reasonable range.

[0093] The total resource requirements of the VNF in the IMS administrative switching core network can be obtained by summing the resource requirements of each network element. When the number of concurrent users in the system is u, the total number of nodes required for normal operation of the core network is N. *The set of network elements corresponding to ≤N single network element nodes is defined as follows: The calculation is performed independently for each resource dimension, and the corresponding formula is as follows:

[0094]

[0095] In the formula, F l This represents the total demand of the VNF for the l-th type of resource. To obtain the simplest configuration for a single core network node, this paper constructs a hybrid networking model with the objective of minimizing the number of network element nodes. The hybrid networking model includes:

[0096] The node set V includes PNFs, VNFs, Layer 2 switches, Layer 3 routers, CE devices, and network management systems;

[0097] Virtual network element instance v is an instance of VNF and is mapped to a virtual machine;

[0098] The edge set E includes virtual links between VNFs and physical links between PNFs;

[0099] The optimization objective of the hybrid networking model is established as follows: under the conditions of satisfying virtual machine resource constraints and traffic conservation constraints, minimize the number of network element nodes contained in the node set V in the core network topology graph G=(V,E).

[0100]

[0101] Virtual machine resource constraints are: for network element node x n ∈X * The virtual network element instance v in the above, the VNF corresponding to it requires no more than the CPU and memory resources allocated by the server to the virtual machine.

[0102] The flow conservation constraint is: through the network element node x n ∈X * The total inflow is equal to the x of the network element node. n ∈X * The sum of the total outflow and the storage change of the server.

[0103] The combination of network element nodes corresponding to the minimum number of network element nodes obtained by solving this problem is the simplest configuration scheme for a single node.

[0104] Based on system fault tolerance and load balancing considerations, the core network should maintain service availability and performance stability during partial failures or load surges. Achieving this requires deploying additional resources while meeting the basic requirements of the VNF (Virtual Network Functions), i.e., adding redundancy to the minimum configuration of a single node. To ensure that single-node failures do not affect service continuity after redundancy is configured, while avoiding performance degradation due to resource overload and supporting dynamic resource expansion during service growth, a hierarchical redundancy configuration method is adopted. First, the redundancy levels of the system are divided, as shown in Table 1.

[0105] Table 1

[0106]

[0107] Let the simplest configuration of a single node obtained by solving formula (10), i.e., the core network basic resource configuration set, be F = {F1(u), F2(u), ..., F...}. l (u),…,F L (u)}, where an integer t≥1 represents the global redundancy level, △ l (u) represents the network element redundancy increment of the l-th resource, calculated as follows:

[0108]

[0109] In the formula, This represents the set of key network elements (such as CSCF) in the simplest configuration of a single node. λ represents the set of non-critical network elements (such as MRFs). n >0 indicates the redundancy coefficient of critical network elements, μ n >0 indicates the redundancy coefficient of non-critical network elements, and satisfies λ n >μ n Then, the single-node redundancy configuration can be represented as:

[0110] Fr l (u)=F l (u)+t·△ l (u), l=1,2,…,L (12);

[0111] The purpose of configuring redundancy for a single node in the core network is to minimize the deployment cost of computing resources, provided that the selected network elements do not exceed the upper limit of VNF resource requirements and the system redundancy meets the safe operating rate. The corresponding optimization problem is as follows:

[0112]

[0113] The first constraint represents the upper limit constraint of resource allocation, and the second constraint represents the safety operation rate constraint. This indicates that when the single-node configuration set is X* The system's safe operating rate when redundancy level is t and concurrent users are u; P * The minimum safe operating rate threshold is set manually based on the system operating environment and historical operating experience.

[0114] Greedy algorithms are an algorithm design paradigm based on local optimal decision-making. The core idea is to gradually construct a global solution to the problem by taking the best local solution in the current state at each step.

[0115] According to the optimization objective (10) and its constraints, when the number of network element nodes reaches its minimum, the resource demand of each network element should be maximized accordingly. Therefore, based on the idea of ​​a greedy algorithm, nodes with larger network element resource demands can be selected step by step, and at the same time compared with the remaining available resources to determine the most suitable node. The variables used in the greedy algorithm and their meanings are shown in Table 2.

[0116] Table 2

[0117]

[0118]

[0119] For the available server resources in the resource pool, the total basic resource requirement is calculated by summing up the network element CPU resources and memory resources separately. The calculation formula is as follows:

[0120]

[0121] The process of implementing the simplest configuration for a single core network node using a greedy algorithm is as follows: Figure 2 As shown.

[0122] Based on the idea of ​​recursive algorithms, the optimization problem (13) can be solved. According to formula (11), the redundancy increment of this problem consists of the redundancy increment of critical network elements and the redundancy increment of non-critical network elements. At the same time, according to the aforementioned core network redundancy hierarchy, the redundancy strategies of non-critical network elements are different. Therefore, a differentiated management strategy is first implemented for the network elements with the simplest configuration of a single node, such as... Figure 3 As shown. Based on the optimization problem (13), the process of implementing the optimal redundancy configuration of the core network through a recursive algorithm is as follows. Figure 4 As shown.

[0123] After obtaining the core network single-node redundancy configuration, to further improve the utilization of deployed redundant resources for IMS service continuity, dual-node redundancy can be configured by combining multi-active instance synchronization technology and dynamic expansion technology. The block diagram of the instance synchronization mechanism is as follows: Figure 5 As shown.

[0124] In this mechanism, session state is implemented using a distributed lock via an ETCD cluster, media streams are transmitted via dual-path RTP / RTCP, and data configuration is performed based on incremental synchronization using rsync. When a single-node redundant configuration is used for network element x... n ∈X * A fault occurred, and x network elements remained. n ∈(XX * When a synchronizable instance exists, the system triggers the above mechanism and adopts... Figure 4 The process configures redundancy to the synchronized single node, thereby completing the network element node switchover and realizing dual-node synchronous redundancy configuration.

[0125] To ensure the core network has sufficient synchronization instances, it is necessary to deploy new VNF instances and adjust service boundaries in real time using dynamic scaling technology. The process of load-aware dynamic scaling algorithm is as follows: Figure 6 As shown.

[0126] In this expansion process, when the VNF capacity utilization rate is below 60%, VNF expansion is performed, creating new VNF instances, migrating users, and verifying the migration results. When the VNF capacity utilization rate is not less than 60%, service cutover is performed, adjusting the PNF-VNF service boundary, performing the cutover within the preset maintenance window, and verifying the cutover results. Furthermore, if expansion fails due to prolonged service interruption, an automatic rollback mechanism is triggered: terminating the newly created VNF instances; switching migrated users back to their original network element nodes; sending an alarm notification to the operations and maintenance management terminal; and recording the corresponding fault codes in the knowledge base.

[0127] The above embodiments are only for illustrating the technical concept and features of the present invention. Their purpose is to enable those skilled in the art to understand the content of the present invention and implement it. They should not be used to limit the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for configuring network element resources in a power IMS virtualization core network, characterized in that, Includes the following steps: S1. Establish a hybrid networking model for the core network, and determine the simplest node configuration and corresponding network element nodes for the core network based on the model; S2. Divide the network element nodes into critical network elements and non-critical network elements, and configure single-node redundancy for the network element nodes based on the division and a preset redundancy coefficient; S3. Configure a synchronous instance for a network element node that has been configured with single-node redundancy to form a dual-node redundancy configuration for the core network.

2. The network element resource allocation method according to claim 1, characterized in that, Step S1 includes: Based on the hybrid networking model, an optimization problem is constructed with the objective of minimizing the number of network element nodes; The optimization problem is solved using a greedy algorithm to obtain the simplest node configuration.

3. The network element resource allocation method according to claim 2, characterized in that, The hybrid networking model includes a set of nodes consisting of physical network elements, virtual network elements, and connecting devices, as well as a set of edges representing the connections between nodes; The constraints of the optimization problem include virtual machine resource constraints and traffic conservation constraints.

4. The network element resource allocation method according to claim 1, characterized in that, Step S2 includes: The basic resource configuration is determined based on the simplified node configuration. Based on the redundancy coefficients of the critical network elements and the redundancy coefficients of the non-critical network elements, calculate the network element redundancy increment. With the goal of minimizing the deployment cost of computing resources, the single-node redundancy configuration is solved by a recursive algorithm, combining the basic resource configuration and the network element redundancy increment.

5. The network element resource allocation method according to claim 1, characterized in that, Step S3 includes: Establish a multi-active instance synchronization mechanism to switch the services of a network element node to a synchronizeable instance on another network element node when a failure occurs.

6. The network element resource allocation method according to claim 5, characterized in that, When the multi-active instance synchronization mechanism is triggered, the following operations are performed: Select a target synchronization instance from the remaining network element nodes; Configure single-node redundancy for the target synchronization instance; The services of the faulty network element node will be switched to the target synchronization instance.

7. The network element resource allocation method according to claim 1, characterized in that, Step S3 further includes: Implement load-aware dynamic capacity expansion, and select to perform VNF expansion or service cutover operations based on the capacity ratio of Virtual Network Functions (VNFs).

8. The network element resource allocation method according to claim 7, characterized in that, When the VNF capacity percentage is lower than the first threshold, a VNF expansion operation is performed; when the VNF capacity percentage is not lower than the first threshold, a service cutover operation is performed.

9. The network element resource allocation method according to claim 7 or 8, characterized in that, The dynamic expansion process also includes an automatic rollback mechanism, which automatically terminates the current operation and switches the service back to its original state when expansion or cutover fails.

10. A network element resource allocation system for a power IMS virtualization core network, characterized in that, The system includes a processing unit configured to perform the network element resource configuration method as described in any one of claims 1 to 9.

Images