Multi-modal network resource block construction method and system

Abstract

The invention relates to the technical field of network resource management, in particular to a multi-mode network resource block construction method and system, and the system comprises a physical resource layer which comprises a physical network and a physical network manager; the virtual resource layer virtualizes physical resources of the physical resource layer, acquires virtual resources, generates resource blocks and deploys and manages the resource blocks through a virtual resource manager, and the virtual resource manager interacts with the physical network manager; the application layer is used for processing a network request, sending a resource demand to the virtual resource layer according to the network request, and scheduling a resource block meeting the resource demand to process a service of the network request; and the request control layer is used for receiving a network request of a user, transmitting the network request to the application layer and acquiring an application service support network request. According to the scheme, the problems of separated management of resources such as calculation, storage and network, low resource scheduling efficiency and the like in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of network resource management, and in particular to a method and system for constructing a multi-modal network resource block. Background Art

[0002] With the rapid development of emerging technologies such as cloud computing, edge computing, and 5G, the demands of enterprises and data centers for network, computing, and storage resources have become increasingly complex. Existing resource management methods are mostly based on a single virtualization technology, which virtualizes and manages physical resources such as computing resources (such as virtual machines, containers), storage resources (such as virtual disks, distributed storage), and network resources (such as virtual networks, virtual switches). Although this traditional virtualization solution has certain elasticity and scalability, it has obvious deficiencies in multi-resource collaborative scheduling, resource utilization efficiency optimization, and business demand adaptation. Especially in multimodal business scenarios, the demands of different businesses for computing, storage, network and other resources are heterogeneous and dynamically changing, and the traditional single resource virtualization model cannot fully meet these demands.

[0003] Although network function virtualization (NFV) and software defined network (SDN) technologies can virtualize and centrally manage network resources, these technologies mostly focus on abstraction and management at the network level, and lack overall optimization management of computing and storage resources. In addition, although existing cloud computing platforms provide a certain degree of flexibility in resource scheduling, their resource management granularity is relatively coarse, usually scheduling resources in units of entire physical nodes or virtual machines, and cannot achieve dynamic and efficient allocation of fine-grained resources, resulting in low resource utilization.

[0004] In summary, the prior art still has the following problems:

[0005] Resource collaborative scheduling problem: In the existing resource scheduling method, various types of resources are managed separately, which makes it difficult to perform unified optimization during scheduling;

[0006] Fine-grained resource scheduling: The resource granularity of traditional virtualization scheduling is large and cannot meet fine-grained business needs;

[0007] Dynamic resource demand adaptation: Existing systems are unable to dynamically perceive and adapt to changes in business demand for computing, storage, and network resources.

[0008] In order to solve the above problems, there is an urgent need for a method and system for constructing a multimodal network resource block, and to build a new resource virtualization model that uniformly abstracts and manages computing, storage, network and other resources, which can not only ensure flexible resource scheduling, but also improve resource utilization efficiency and business adaptation capabilities. Summary of the invention

[0009] One of the purposes of the present invention is to provide a system for constructing a multimodal network resource block. In view of the problems of separate management of computing, storage, network and other resources and low resource scheduling efficiency in the prior art, computing, storage, network and other resources are integrated into resource blocks to form a resource unit that can be flexibly scheduled and dynamically expanded, thereby improving resource utilization, reducing resource fragmentation, and being able to adapt to the heterogeneous resource requirements of different businesses, so as to achieve efficient collaborative management of resources through unified abstraction and dynamic scheduling of multiple resources.

[0010] The present invention provides a basic solution 1: a system for constructing a multimodal network resource block, comprising:

[0011] Physical resource layer, including: physical network and physical network manager;

[0012] The virtual resource layer virtualizes the physical resources of the physical resource layer, obtains virtual resources, generates resource blocks, and allocates and manages them through the virtual resource manager, which also interacts with the physical network manager; the physical resources are the physical network and the physical network manager; the virtual resources include computing resources, storage resources, and network resources;

[0013] The application layer is used to process network requests, send resource requirements to the virtual resource layer according to network requests, and schedule resource blocks that meet the resource requirements to process the network request services;

[0014] The request control layer is used to receive the user's network request and pass the network request to the application layer to obtain application service support for the network request.

[0015] Furthermore, the virtualizing the physical resources of the physical resource layer to obtain virtual resources includes:

[0016] To model physical assets:

[0017] Abstract the physical network of the physical resource layer into an undirected weighted graph Indicates that N s is the set of all underlying physical nodes, L s is a link set consisting of physical links, is the attribute set contained in the physical node. Assume that each node has n attributes, P1, P2, ... Pn. These attributes represent the types of resources that the node can provide, such as node CPU resources, memory resource capacity, storage resource capacity, etc. It is a set of attributes contained in the physical link;

[0018] To model virtual resources:

[0019] Abstract the virtual network as an undirected weighted graph Indicates that N vrepresents the set of virtual nodes, L v represents a set of virtual links, Represents the attribute set contained in the virtual node. Assume that each node has n attributes, P1, P2, ... Pn, which represent the resource types required by the node; Indicates the demand of the virtual link for the underlying physical link resources;

[0020] Mapping virtual nodes and virtual links to physical nodes and physical links;

[0021] By integrating and scheduling virtual resources, resource blocks that can be used by the business are generated.

[0022] Furthermore, different virtual nodes can be mapped to the same physical node, and different virtual links can also be mapped to the same physical link.

[0023] Furthermore, the constraints for resource block construction are:

[0024] The resource blocks generated on the entire physical network must meet the preset conditions. For each physical node, the computing and storage resources opened by all resource blocks cannot be greater than the computing and storage resources that the node can provide. For each physical link, the network resources opened cannot be greater than the resources that the link can provide. Specifically, they include:

[0025] If the physical node is N, the physical link is L, and the resource block is B;

[0026] For each physical node, let the computing and storage resources that physical node i can provide be in Represents resources with computing and storage attributes, such as CPU resources, GPU resources, memory resources, storage resources, etc.

[0027] For each link, let the network resources provided by link k be in Represented as a resource with network attributes;

[0028] If in a certain network state, there are m resource blocks, u physical nodes, and v physical links, let the computing and storage resources required by resource block j at physical node i be The network resources required at link k are For each physical node, the following constraints are established for computing and storage resources:

[0029]

[0030] For each physical link, the following constraints are established for network resources:

[0031]

[0032] Furthermore, the resource block further comprises: a control and monitoring module;

[0033] The control and monitoring module is the core management unit of the resource block operation, which is used to monitor, schedule and optimize various resources in the resource block.

[0034] Furthermore, the resource block also includes: an API interface and an integration module;

[0035] API interface and integration module are used for resource blocks to provide standardized API interfaces to the outside world.

[0036] Furthermore, Kubernetes is used as a container orchestration tool to generate and manage resource blocks.

[0037] The second purpose of the present invention is to provide a method for constructing a multimodal network resource block. In view of the problems of separate management of computing, storage, network and other resources and low resource scheduling efficiency in the prior art, by integrating computing, storage, network and other resources into resource blocks, a resource unit that can be flexibly scheduled and dynamically expanded is formed, which improves resource utilization, reduces resource fragmentation, and can adapt to the heterogeneous resource requirements of different businesses, so as to achieve efficient collaborative management of resources through unified abstraction and dynamic scheduling of multiple resources.

[0038] The present invention provides a second basic solution: a method for constructing a multimodal network resource block, and a system for constructing the multimodal network resource block, comprising:

[0039] S1. Set resource attributes;

[0040] S2, perception and quantification of physical resources;

[0041] S3, virtualize physical resources to obtain virtual resources;

[0042] S4. Generate resource blocks based on virtual resources, and dynamically expand, monitor and optimize resource blocks.

[0043] Beneficial effects: This solution aims at the problem of resource collaborative scheduling. The virtual resource layer virtualizes the physical resources of the physical resource layer, obtains virtual resources, generates resource blocks, and manages and schedules multiple resources in a unified manner through the design of resource blocks to achieve efficient collaboration.

[0044] Fine-grained resource scheduling is carried out through the virtual resource manager, which also interacts with the physical network manager. Resource blocks can be scheduled in fine-grained units to improve resource utilization efficiency.

[0045] In response to dynamic resource demand adaptation, the request control layer receives the user's network request and passes the network request to the application layer to obtain application services to support the network request. The application layer processes the network request and sends resource requirements to the virtual resource layer based on the network request. The resource blocks that meet the resource requirements are scheduled to process the network request business. The virtual resource layer will allocate and manage through the virtual resource manager. Through the dynamic adjustment mechanism of resource blocks, the system can respond to business needs in a timely manner and dynamically expand and shrink resources.

[0046] This solution addresses the problems of separate management of computing, storage, network and other resources and low resource scheduling efficiency in the existing technology. By integrating computing, storage, network and other resources into resource blocks, a flexibly schedulable and dynamically expandable resource unit is formed to improve resource utilization, reduce resource fragmentation, and adapt to the heterogeneous resource requirements of different businesses. Through unified abstraction and dynamic scheduling of multiple resources, efficient collaborative management of resources can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic diagram of a system embodiment for constructing a multimodal network resource block according to the present invention;

[0048] Figure 2 A schematic diagram of the physical network and virtual network topology in a system embodiment of the present invention for constructing a multimodal network resource block. DETAILED DESCRIPTION

[0049] The following is further described in detail through specific implementation methods:

[0050] The embodiment is basically as shown in the attached Figure 1 As shown:

[0051] A system for constructing a multimodal network resource block, comprising:

[0052] The physical resource layer includes: physical network and physical network manager; the physical network is a network of multiple data centers in different geographical locations, and each physical network has an independent physical network manager;

[0053] The virtual resource layer virtualizes the network resources (physical resources) of the physical resource layer, obtains virtual resources, generates resource blocks, and allocates and manages them through the virtual resource manager, which also interacts with the physical network manager; the physical resources are the physical network and the physical network manager; the virtual resources include but are not limited to: computing resources, storage, and network bandwidth; virtual resources exist in a virtual form for allocation and management by the virtual resource manager, which is not only responsible for the integration and scheduling of virtual resources, but also interacts with the underlying physical network manager to ensure that the flexibility and dynamism of network resources are maximized;

[0054] The application layer is used to process network requests, send resource requirements to the lower layer (i.e., virtual resource layer) based on network requests, and schedule resource blocks that meet the resource requirements to process the services requested by the network;

[0055] The request control layer is used to receive the user's network request and pass the network request to the next layer (i.e., the application layer) to obtain application services to support the network request;

[0056] The network resources (physical resources) of the physical resource layer are virtualized to obtain virtual resources, that is, the physical resources (i.e., the network resources of the physical resource layer) are virtualized. The principle is: when building a model for the entire network, the physical resources and virtual resources are modeled, and the complex network topology is represented by data. In this way, the network structure and resource allocation can be better understood, resource utilization can be optimized, and flexible deployment and management of network functions can be achieved. This modeling method provides an important mathematical basis for the construction of network resource blocks, helping to achieve the effective allocation and utilization of network resources.

[0057] like Figure 2 As shown in the figure, a visual representation of the topology of the physical network and the virtual network is given, and the result of mapping a virtual network to a physical network is given; Figure (b) is a virtual network, which contains three virtual nodes (a, b, c) and three virtual links connecting these three nodes. The bandwidth resources required by the virtual link are marked next to the virtual network link, and the physical resources required by the virtual node (such as CPU resources, memory resources) are marked in brackets next to the virtual node; Figure (a) is a physical network, which consists of 6 physical nodes and 8 physical links. The solid line connecting each physical node represents the physical link, and the value in brackets next to the physical node (A, B, C, D, E, F) represents the physical resources that the physical node can provide (such as CPU resources, memory resources), and the value in brackets next to the physical link is the bandwidth resources that this physical link can provide; the dotted line connection represents the mapping relationship between the virtual nodes in the virtual network and the physical nodes in the physical network;

[0058] Specifically, the network resources of the physical resource layer are virtualized to obtain virtual resources, including:

[0059] Physical asset modeling

[0060] Abstract the underlying physical network (physical resource layer) into an undirected weighted graph Indicates that N s is the set of all underlying physical nodes, L s is a link set consisting of physical links, is the attribute set contained in the physical node. Assume that each node has n attributes, P1, P2, ... Pn. These attributes represent the types of resources that the node can provide, such as node CPU resources, memory resource capacity, storage resource capacity, etc. It is a set of attributes contained in the physical link (such as bandwidth resources);

[0061] Virtual Resource Modeling

[0062] Abstract the virtual network as an undirected weighted graph Indicates that N v represents the set of virtual nodes, L v represents a set of virtual links, Represents the attribute set contained in the virtual node. Assume that each node has n attributes, P1, P2, ... Pn. These attributes represent the resource types required by the node, such as node CPU resources, memory resource capacity, storage resource capacity, etc. Indicates the demand of the virtual link for the underlying physical link resources (such as bandwidth resources).

[0063] Map virtual nodes and virtual links to physical nodes and physical links; different virtual nodes can be mapped to the same physical node, and different virtual links can also be mapped to the same physical link.

[0064] By integrating and scheduling virtual resources, resource blocks available for business use are generated. Specifically, physical resources are abstracted into logical resource pools through virtualization technology to facilitate business scheduling and unified management. Logical resource pools provide basic resources for resource blocks, and resource blocks are allocation units of logical resource pools, which can be flexibly configured and scheduled according to business needs.

[0065] Constraints for resource block construction

[0066] The resource blocks generated on the entire physical network must meet the preset conditions. That is, for each physical node, the computing and storage resources opened by all resource blocks cannot be greater than the computing and storage resources that the node can provide. For each physical link, the network resources opened cannot be greater than the resources that the link can provide. Specifically, they include:

[0067] If the physical node is N, the physical link is L, and the resource block is B;

[0068] For each physical node, let the computing and storage resources that physical node i can provide be in Represents resources with computing and storage attributes, such as CPU resources, GPU resources, memory resources, storage resources, etc.

[0069] For each link, let the network resources provided by link k be in Represented as resources with network attributes, such as bandwidth;

[0070] If in a certain network state, there are m resource blocks, u physical nodes, and v physical links, let the computing and storage resources required by resource block j at physical node i be The network resources required at link k are Considering each physical node, the constraints established for computing and storage resources are:

[0071]

[0072] Considering each physical link, the constraints established for network resources are:

[0073]

[0074] In addition, the resource block also includes: control and monitoring module and API interface and integration module;

[0075] The control and monitoring module is the core management unit of the resource block operation, which is used to monitor, schedule and optimize various resources in the resource block;

[0076] The API interface and integration module are used for the resource block to provide a standardized API interface to the outside world, so that external business systems can easily call various functions in the resource block.

[0077] Specifically, the resource block acts as a logical abstraction layer to encapsulate the underlying physical resources for dynamic invocation according to business needs. Each resource block consists of computing resources, storage resources, network resources, control and monitoring modules, etc., ultimately forming a unified resource scheduling unit that can adapt to the multimodal needs of different types of businesses. The resource block interacts with the upper-layer business logic through APIs or controllers to achieve intelligent scheduling and optimal resource allocation:

[0078] Computing resources: one of the core components of the resource block, mainly providing processing capabilities; computing resources are abstracted and encapsulated through virtualization and containerization technologies to support different types of computing tasks;

[0079] Storage resources: Storage resources are responsible for managing and providing persistent data storage, supporting multiple types of storage services to meet different business needs;

[0080] Network resources: Network resources provide interconnection capabilities for computing and storage modules in resource blocks, ensuring efficient transmission of data and tasks. Network resources rely on virtualization technology to achieve on-demand allocation and dynamic adjustment.

[0081] Control and monitoring module: The control and monitoring module is the core management unit of the resource block operation, responsible for monitoring, scheduling and optimizing various resources in the resource block;

[0082] API interface and integration module: The resource block system provides a standardized API interface to the outside world, allowing external business systems to easily call various functions in the resource block.

[0083] The specific implementation of each component of the resource block in this solution is to use Kubernetes as a container orchestration tool, abstract different physical resources into a unified logical resource pool, and use its plug-in mechanism to achieve the collaborative work of different components of the resource block.

[0084] Through the tools and technologies of the cloud-native architecture, the abstraction and management of computing, storage, and network resources are unified, including:

[0085] Unified management of computing resources

[0086] Virtualization and unified management of computing resources are achieved through containerization, which abstracts computing resources into lightweight, isolated operating units and allows these containers to be scheduled on multiple physical servers;

[0087] The container orchestration function of Kubernetes is used to integrate the CPU and GPU resources of multiple servers and manage them in the form of logical resource blocks. The computing tasks are packaged into containers through Pods in Kubernetes, and each container corresponds to a part of computing resources (CPU, memory, GPU).

[0088] Kubernetes provides a mechanism for resource requests and resource limits to ensure that each container uses only the computing resources it needs, so that computing resources can be scheduled in a fine-grained manner while ensuring isolation within resource blocks.

[0089] Unified management of storage resources

[0090] Unified management is achieved using the Container Storage Interface (CSI) provided by Kubernetes. CSI is a standardized storage plug-in mechanism that allows different storage backends to be seamlessly integrated with Kubernetes.

[0091] Ceph is used as the storage backend and managed through the Rook Operator of Kubernetes. Rook integrates distributed storage systems such as Ceph into Kubernetes, allowing high-availability and elastically scalable storage services to be provided in resource blocks. Each storage volume is the storage part of a resource block and is allocated to the business on demand, ensuring flexible allocation of storage resources across physical nodes.

[0092] The expansion of storage resources is achieved through the dynamic volume scheduling mechanism of Kubernetes; when the storage demand of a business request increases, Kubernetes automatically creates a new persistent volume according to the storage definition policy and binds it to the corresponding resource block to ensure the dynamic scalability of storage resources.

[0093] Unified management of network resources

[0094] The virtualization and management of network resources are implemented through the Kubernetes network plug-in (CNI, Container Network Interface); CNI is a standardized network interface plug-in in Kubernetes, which allocates the required network resources to each resource block and ensures network isolation and security.

[0095] Unified management and scheduling strategy

[0096] Use Kubernetes as a unified management platform for resource blocks, and collaboratively manage computing, storage, and network resources through its scheduler and controller; the Kubernetes scheduler dynamically schedules various resources in the resource block based on business requirements in business requests.

[0097] The operating status of resource blocks is monitored in real time through monitoring tools such as Prometheus. Through the combination of Prometheus and Kubernetes, information such as CPU usage, memory consumption, storage occupancy, and network traffic within the resource blocks can be monitored in a fine-grained manner, and dynamic adjustments can be made based on the monitoring data.

[0098] Combined with Kubernetes' horizontal pod autoscaler (HPA) and vertical pod autoscaler (VPA), the scale of resource blocks is dynamically adjusted according to real-time business load to ensure maximum resource utilization efficiency.

[0099] This embodiment also provides a method for constructing a multimodal network resource block, and a system using the above-mentioned construction of a multimodal network resource block includes:

[0100] S1. Set resource attributes;

[0101] S2, perception and quantification of physical resources;

[0102] S3, virtualize physical resources to obtain virtual resources;

[0103] S4. Generate resource blocks based on virtual resources, and dynamically expand, monitor and optimize resource blocks.

[0104] In this embodiment, taking the compatibility of two modes as an example, implementation plans are given respectively:

[0105] 1. Resource block construction under IP mode

[0106] S1. Set resource attributes: Set resource attributes under IP mode, including but not limited to: CPU performance, GPU performance, memory capacity, storage IO performance, hard disk capacity, and network bandwidth;

[0107] S2, perception and quantification of physical resources;

[0108] The perception and quantification of physical resources is the basis for ensuring the effective implementation of resource virtualization. The specific steps are:

[0109] Resource awareness tools: Use container orchestration platforms such as Kubernetes to perceive the resource usage of all nodes in the cluster through its built-in Node Exporter module; Node Exporter can capture the real-time status of the node's CPU, memory, network, and storage, and collect and quantify data through Prometheus;

[0110] Resource quantification standards: Set resource quantification standards and abstract each physical resource into different types of resource block components. For example, the physical CPU is quantified as the number of cores, the memory is in GB, the storage resources are quantified in IOPS and capacity, and the network resources are quantified according to the bandwidth (such as Gbps).

[0111] S3, virtualize physical resources to obtain virtual resources;

[0112] Through virtualization technology, the underlying physical resources are abstracted into logical resource pools to facilitate the scheduling and unified management of upper-layer services. The specific steps are as follows:

[0113] Computing resource virtualization: Through container technology (such as Docker), the CPU and memory resources of physical servers are virtualized into lightweight, dynamically allocated computing resource pools; Kubernetes, as an orchestrator, manages these containerized resources and allocates computing tasks to appropriate physical nodes through its scheduler;

[0114] Storage resource virtualization: Through distributed storage systems (such as Ceph), multiple physical storage nodes are virtualized into a unified storage pool. With the help of Kubernetes' CSI (Container Storage Interface) plug-in, storage resources are created and allocated on demand to ensure data persistence and high availability.

[0115] Network resource virtualization: Use Kubernetes network plug-ins (such as CNI) to virtualize the underlying physical network into an independent virtual network to support data transmission between components in the resource block; at the same time, dynamic scheduling and optimization of network resources are achieved through SDN (software defined network) technology;

[0116] S4. Generate resource blocks based on virtual resources, and dynamically expand, monitor and optimize resource blocks;

[0117] By integrating and scheduling virtualized resources, resource blocks that can be used by the business are finally generated. The specific steps are as follows:

[0118] Generation of resource blocks: After the system virtualizes the perceived and quantified physical resources, it combines computing resources, storage resources, network resources, etc. on demand through the Kubernetes scheduler to generate resource blocks that can be used by the business. Each resource block is a logical unit containing multiple resources and can be scheduled independently.

[0119] Dynamic resource block expansion: When business needs change, resource blocks can be dynamically expanded according to changes in business load. Through Kubernetes's HPA (Horizontal Pod Autoscaler) and VPA (Vertical Pod Autoscaler), the system can dynamically expand the number of computing resources or adjust the resources occupied by existing computing tasks to ensure the dynamic elasticity of resource blocks;

[0120] Resource block monitoring and optimization: The system uses monitoring tools such as Prometheus to monitor resource blocks in real time, including CPU usage, memory usage, network traffic, and storage IOPS.

[0121] This solution can effectively perceive the usage of physical resources, and uniformly abstract and manage these resources through virtualization technology, ultimately generating resource blocks containing computing, storage, and network resources, thus achieving fine-grained scheduling and dynamic expansion of resources.

[0122] 2. Resource construction in satellite mode

[0123] The steps for constructing resource blocks are the same as those in the above IP mode. The difference is that the resource block construction in different modes is mainly reflected in the differences in some resource attributes. In the satellite mode, the resource attributes are set as: ground station capability (unit: GFLOPS), storage capacity (unit: GB), and satellite communication bandwidth (unit: Mbps).

[0124] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme is not described in detail here. The ordinary technicians in the relevant field know all the common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, which will not affect the effect of the implementation of the present invention and the practicality of the patent. The protection scope required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A system for constructing a multimodal network resource block, characterized in that: include: Physical resource layer, including: physical network and physical network manager; The virtual resource layer virtualizes the physical resources of the physical resource layer, obtains virtual resources, generates resource blocks, and allocates and manages them through the virtual resource manager, which also interacts with the physical network manager; the physical resources are the physical network and the physical network manager; the virtual resources include computing resources, storage resources, and network resources; The application layer is used to process network requests, send resource requirements to the virtual resource layer according to network requests, and schedule resource blocks that meet the resource requirements to process the network request services; The request control layer is used to receive the user's network request and pass the network request to the application layer to obtain application service support for the network request.

2. The system for constructing a multimodal network resource block according to claim 1, characterized in that: The step of virtualizing the physical resources of the physical resource layer to obtain virtual resources includes: To model physical assets: Abstract the physical network of the physical resource layer into an undirected weighted graph Indicates that N s is the set of all underlying physical nodes, L s is a link set consisting of physical links, is the attribute set contained in the physical node. Assume that each node has n attributes, P1, P2, ... Pn. These attributes represent the types of resources that the node can provide, such as node CPU resources, memory resource capacity, storage resource capacity, etc. It is a set of attributes contained in the physical link; To model virtual resources: Abstract the virtual network as an undirected weighted graph Indicates that N v represents the set of virtual nodes, L v Represents a set of virtual links. Represents the attribute set contained in the virtual node. Assume that each node has n attributes, P1, P2, ... Pn, which represent the resource types required by the node; Indicates the demand of virtual links for underlying physical link resources; Mapping virtual nodes and virtual links to physical nodes and physical links; By integrating and scheduling virtual resources, resource blocks that can be used by the business are generated.

3. The system for constructing a multimodal network resource block according to claim 2, characterized in that: Different virtual nodes can be mapped to the same physical node, and different virtual links can also be mapped to the same physical link.

4. The system for constructing a multimodal network resource block according to claim 2, characterized in that: The constraints for resource block construction are: The resource blocks generated on the entire physical network must meet the preset conditions. For each physical node, the computing and storage resources opened by all resource blocks cannot be greater than the computing and storage resources that the node can provide. For each physical link, the network resources opened cannot be greater than the resources that the link can provide. Specifically, they include: If the physical node is N, the physical link is L, and the resource block is B; For each physical node, let the computing and storage resources that physical node i can provide be in Represents resources with computing and storage attributes, such as CPU resources, GPU resources, memory resources, storage resources, etc. For each link, let the network resources provided by link k be in Represented as a resource with network attributes; If in a certain network state, there are m resource blocks, u physical nodes, and v physical links, let the computing and storage resources required by resource block j at physical node i be The network resources required at link k are For each physical node, the following constraints are established for computing and storage resources: For each physical link, the following constraints are established for network resources:

5. The system for constructing a multimodal network resource block according to claim 1, characterized in that: The resource block also includes: a control and monitoring module; The control and monitoring module is the core management unit of the resource block operation, which is used to monitor, schedule and optimize various resources in the resource block.

6. The system for constructing a multimodal network resource block according to claim 1, characterized in that: The resource block also includes: an API interface and an integration module; API interface and integration module are used for resource blocks to provide standardized API interfaces to the outside world.

7. The system for constructing a multimodal network resource block according to claim 1, characterized in that: Kubernetes is used as a container orchestration tool to generate and manage resource blocks.

8. A method for constructing a multimodal network resource block, characterized in that: A system for constructing a multimodal network resource block according to any one of claims 1 to 7, comprising: S1. Set resource attributes; S2, perception and quantification of physical resources; S3, virtualize physical resources to obtain virtual resources; S4. Generate resource blocks based on virtual resources, and dynamically expand, monitor and optimize resource blocks.

Images