CN114615180A

CN114615180A - Calculation force network system, calculation force calling method and device

Info

Publication number: CN114615180A
Application number: CN202210239786.7A
Authority: CN
Inventors: 蔡慧; 石磊; 姚怡东; 周海涛
Original assignee: Alibaba Damo Institute Hangzhou Technology Co Ltd
Current assignee: Alibaba Damo Institute Hangzhou Technology Co Ltd
Priority date: 2022-03-09
Filing date: 2022-03-09
Publication date: 2022-06-10

Abstract

The embodiment of the specification provides a computing power network system, a computing power calling method and a computing power calling device, wherein the computing power network system comprises: the system comprises a calculation force service module, a calculation force control module and at least two calculation force nodes, wherein the at least two calculation force nodes comprise a calculation force routing layer; the computing power routing layer is configured to generate a computing power routing table based on the obtained computing power information and upload the computing power routing table to the computing power control module, wherein the computing power routing table records computing power capability parameters and network capability parameters of a local computing power node and a neighbor computing power node; the computational force control module is configured to determine a target computational force policy based on the computational force routing table, wherein the target computational force policy comprises a target computational force destination address and target routing path information.

Description

Calculation force network system, calculation force calling method and device

Technical Field

The embodiment of the specification relates to the technical field of computers, in particular to a computing power network system and a computing power calling method.

Background

With the deepening of digital transformation in production and operation of various industries, various industry terminals generate massive original data, so that a great deal of calculation is needed to process; the current single-terminal equipment cannot be increased in proportion along with the increase of the number of multi-core chips, so that the computational power resources of the single-terminal equipment are limited, the service for all services at a specific time point cannot be provided, and meanwhile, the response time of service layer scheduling is in the second level, and the application with low delay and high service continuity requirements cannot be met.

Disclosure of Invention

In view of this, the embodiments of the present specification provide a computing power network system. One or more embodiments of the present specification also relate to an computing power calling method and a computing power calling apparatus, a computing device, a computer-readable storage medium, and a computer program, so as to solve the technical disadvantages in the prior art.

According to a first aspect of embodiments herein, there is provided a computing power network system, including: the computing power system comprises a computing power service module, a computing power control module and at least two computing power nodes, wherein the at least two computing power nodes comprise a computing power routing layer;

the computing power routing layer is configured to generate a computing power routing table based on the obtained computing power information and upload the computing power routing table to the computing power control module, wherein the computing power routing table records computing power capability parameters and network capability parameters of a local computing power node and a neighbor computing power node;

the computational force control module is configured to determine a target computational force policy based on the computational force routing table, wherein the target computational force policy comprises a target computational force destination address and target routing path information.

According to a second aspect of the embodiments of the present specification, there is provided a computing power calling method applied to a computing power network system, the computing power network system including a computing power service module, a computing power control module and at least two computing power nodes,

the computing power service module generates a target computing power calling instruction based on computing power information carried in a computing power calling request sent by a target user;

the calculation force control module acquires calculation force routing tables generated by the at least two calculation force nodes based on the target calculation force calling instruction, and determines a target calculation force strategy based on the calculation force routing table of each calculation force node, wherein the target calculation force strategy comprises a target calculation force capability parameter corresponding to the calculation force information and a target network capability parameter;

the calculation force control module determines a routing address of a target calculation force node from a preset routing address table based on the target calculation force strategy and calls a calculation resource corresponding to the calculation force information for the target user based on the routing address;

the preset routing address table comprises routing node marks and routing addresses of all the computation force nodes.

According to a third aspect of the embodiments of the present specification, there is provided an computing power invoking device applied to a computing power network system, where the computing power network system includes a computing power service module, a computing power control module, and at least two computing power nodes,

the computing power service module is configured to generate a target computing power calling instruction based on computing power information carried in a computing power calling request sent by a target user;

the computing force control module is configured to obtain computing force routing tables generated by the at least two computing force nodes based on the target computing force calling instruction, and determine a target computing force strategy based on the computing force routing table of each computing force node, wherein the target computing force strategy comprises a target computing force capability parameter corresponding to the computing force information and a target network capability parameter;

the computing force control module is further configured to determine a routing address of a target computing force node from a preset routing address table based on the target computing force strategy, and call a computing resource corresponding to the computing force information for the target user based on the routing address;

According to a fourth aspect of embodiments herein, there is provided a computing device comprising:

a memory and a processor;

the memory is configured to store computer-executable instructions and the processor is configured to execute the computer-executable instructions, which when executed by the processor implement the steps of the computational power calling method described above.

According to a fifth aspect of embodiments herein, there is provided a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the computational power calling method described above.

According to a sixth aspect of embodiments herein, there is provided a computer program, wherein the computer program, when executed in a computer, causes the computer to perform the steps of the above-mentioned algorithm calling method.

One embodiment of the present specification provides a computing power network system, including: the system comprises a calculation force service module, a calculation force control module and at least two calculation force nodes, wherein the at least two calculation force nodes comprise a calculation force routing layer; the computing power routing layer is configured to generate a computing power routing table based on the obtained computing power information and upload the computing power routing table to the computing power control module, wherein the computing power routing table records computing power capability parameters and network capability parameters of a local computing power node and a neighbor computing power node; the computational force control module is configured to determine a target computational force policy based on the computational force routing table, wherein the target computational force policy comprises a target computational force destination address and target routing path information.

Specifically, in the embodiment of the present specification, a computational power network system is established, and a plurality of computational power nodes are deployed, so as to implement selection between the plurality of computational power nodes through a computational power policy, thereby avoiding a situation that a service cannot be provided due to provision of computational power by only a single device terminal, and meanwhile, each computational power node includes a computational power routing layer to generate a computational power routing table and report the computational power routing table to a computational power control module, which facilitates the computational power control module to determine a target computational power policy according to the computational power routing table reported by each computational power node, thereby implementing selection of a more appropriate computational power node among the plurality of computational power nodes to provide computational power.

Drawings

FIG. 1 is a schematic diagram of a system architecture of a computational power network system provided in one embodiment of the present disclosure;

fig. 2 is a schematic diagram illustrating a service data packet forwarding flow of a computing power forwarding plane in a computing power invoking system according to an embodiment of the present disclosure;

fig. 3 is a schematic flowchart of reporting computing power routing information of a computing power invoking system according to an embodiment of the present disclosure;

FIG. 4 is a schematic flow chart illustrating computation force route generation in a computation force invocation system according to an embodiment of the present disclosure;

FIG. 5 is a flow diagram of a method for computing power invocation provided by an embodiment of the present description;

FIG. 6 is a process diagram of a computing power calling method according to an embodiment of the present disclosure;

FIG. 7 is a schematic structural diagram of an algorithm invocation device provided in an embodiment of the present specification;

fig. 8 is a block diagram of a computing device according to an embodiment of the present disclosure.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present description. This description may be implemented in many ways other than those specifically set forth herein, and those skilled in the art will appreciate that the present description is susceptible to similar generalizations without departing from the scope of the description, and thus is not limited to the specific implementations disclosed below.

The terminology used in the description of the one or more embodiments is for the purpose of describing the particular embodiments only and is not intended to be limiting of the description of the one or more embodiments. As used in one or more embodiments of the present specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present specification refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It will be understood that, although the terms first, second, etc. may be used herein in one or more embodiments to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first can also be referred to as a second and, similarly, a second can also be referred to as a first without departing from the scope of one or more embodiments of the present description. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.

First, the noun terms to which one or more embodiments of the present specification relate are explained.

Calculating the strength: the method is a general term for data processing capacity and services, is composed of upper-layer services formed by various chips, components and packages, has diversity in computing power, and is a development foundation for technologies such as cloud computing, edge computing, big data and artificial intelligence.

Computing force network: a novel network technical scheme needs high cooperation of a network and calculation, a calculation unit and calculation capacity are embedded into the network, the tidal effect of calculation capacity is relieved by utilizing ubiquitous idle calculation capacity in a network calculation mode, and the utilization rate of calculation resources is improved.

Cloud network fusion: the technology of introducing the network into the cloud computing and the technology of introducing the cloud computing into the communication network are referred to.

And (3) network computing cooperation: through cooperative sensing of the service, the computing power resource and the network resource, the service is dispatched to a proper node according to needs, unified arrangement, unified operation and maintenance and unified optimization of the computing network resource are realized, and finally, computing network common-bounce and common-shrinkage are realized.

Integrating calculation and network: on the basis of cloud network fusion, the method emphasizes the combination of changes of future business forms, continuously promotes research and development on 3 layers of cloud, network and core, and realizes application deployment matching calculation, network forwarding perception calculation and chip capability enhancement calculation.

And (3) edge calculation: the network and the calculation are separated, and the network is firstly transmitted through the network and then processed by the edge data center.

With the deepening of digital transformation in production and operation of various industries, terminals of various industries generate massive original data. In digital production, data from a large number of IoT and video capture signals will require at least giga of network access capabilities. In the holography stage, if information such as human appearance, touch, smell, taste and the like needs to be transmitted holographically, uncompressed data needs 4Tbps, and even after compression, at least one hundred G of access bandwidth is needed. Data processing requires a networked computational new infrastructure: the massive data generated by digital transformation needs to be processed with a great deal of computing power urgently, and the computing power demand of various industries is increased at a high speed. Furthermore, the development of the computing power network is distributed from a centralized direction, at present, the computing power of single-terminal equipment cannot be increased in proportion with the increase of the number of the cores of the multi-core chip, so that the number of the cores of the multi-core chip approaches to the upper limit when the number of the cores of the multi-core chip is 128 cores; the cloud data center is limited by transmission bandwidth cost and time delay, and cannot meet the performance requirement of future service application. Therefore, edge computing power which is flexible and low in cost becomes a key ring for supporting the processing of mass data in the intelligent society, computing is driven to move from the cloud end to the edge side close to a data source, and distributed computing power resources in a network are formed, so that a cloud-edge-end ubiquitous deployment architecture is inevitably constructed for future computing power, and computing power requirements brought by the digital development of the society are met.

In addition, the computational resources of a single current site are limited, and it cannot be guaranteed that the SLA (Service-level agreement) required for all services is provided at a specific time point, and meanwhile, the response time of Service layer scheduling is at the second level, and it cannot meet the application with low latency and high Service continuity requirements, so that it is necessary that computational nodes perform cooperative work, a computational task is scheduled to a better computational node for processing, and the computational task is scheduled to the better computational node for processing.

Based on this, the computational power network system provided in the embodiment of the present specification implements dynamic sensing and on-demand scheduling of the full-network computational power resources through the technical characteristic architectures such as computational power sensing, control, and scheduling, and implements full-network computational power state propagation and application scheduling through two deployment manners, namely, centralized deployment and distributed deployment.

In the present specification, a computing power network system is provided, and the present specification relates to a computing power calling method and a computing power calling apparatus, a computing device, and a computer-readable storage medium, which are described in detail in the following embodiments one by one.

Referring to fig. 1, fig. 1 is a schematic diagram illustrating a system architecture of a computational power network system according to an embodiment of the present disclosure.

The computational network system 100 of fig. 1 may include a computational service module 102, a computational control module 104, and at least two computational nodes 106, wherein the at least two computational nodes include a computational routing layer 1062.

Specifically, the computation routing layer 1062 is configured to generate a computation routing table based on the obtained computation information, and upload the computation routing table to the computation force control module 104, where the computation routing table records computation capability parameters and network capability parameters of a local computation node and a neighboring computation node; the computational force control module 104 is configured to determine a target computational force policy based on the computational force routing table, wherein the target computational force policy includes a target computational force destination address and target routing path information.

The computing power service module is used for realizing external publishing of computing power and providing computing power service for external application in an interface calling mode; the calculation force control module can be understood that a global calculation force control layer and a flexible and quick edge calculation force control layer are cooperated to form a distributed and layered calculation force control system, and the global calculation force control layer is cooperated with a distributed calculation force control layer and a universal terminal calculation force function to realize end-to-end internal calculation force control.

The computing power routing layer can be understood as the core of the computing power sensing network, and flexibly schedules the services of the layer to different computing resource nodes according to the needs based on the abstracted computing network resources and comprehensively considering the network conditions and the computing resource conditions.

In practical application, the computation routing layer is deployed in forwarding equipment of a computation system framework, a computation routing table between I P layers and an application layer is used for constructing a computation model of a local device or a local cluster, resource state information of a neighbor computation routing node is obtained through a cooperative computation force control module, and a target computation strategy is determined to achieve optimal scheduling of services.

It should be noted that the target computation power policy includes a target computation power destination address and target routing path information, where the target computation power destination address may be understood as a routing address of a target computation power node, and the target routing path information may be understood as node information of a routing path forwarded by communication when a computation power requester establishes a communication connection with the target computation power node, for example, if the routing address of the target computation power node is F, and the routing address of the local computation power requester is a, then the target routing path information may be a-C-D-B-E-F (letters indicate the routing address of each computation node).

Further, the calculation power routing layer generates a calculation power routing table according to the calculation power information acquired in real time, wherein the calculation power routing table comprises calculation power capability parameters and network capability parameters of a local calculation power node and a neighboring calculation power node, and the calculation power parameters comprise calculation type parameters and calculation performance evaluation parameters, such as CPU/GPU calculation power, calculation time delay and other extensible parameters; the network capability parameters comprise network forwarding bandwidth, network transmission delay and network forwarding path; and solving the optimal execution node through weighting the sum of the calculation parameters and the network parameters, so as to realize the calculation route forwarding of the service.

In the computational power network system provided in the embodiment of the present specification, a computational power routing table generated by a computational power routing layer in the computational power node is used to synthesize the computational power and the routing capability to obtain a better computational power node, so that the transmission delay and the computational delay of a routing communication path are both greatly reduced to meet the performance requirements of applications.

Further, the computing power service module comprises a computing power calling layer and a computing power template layer,

the computing power calling layer is configured to receive a computing power calling request sent by a target user and send initial computing power information carried in the computing power calling request to the computing power template layer;

the computing power template layer is configured to determine target computing power information based on the initial computing power information and a preset computing power template, and generate a target computing power calling instruction based on the target computing power information.

The computing power calling layer can be understood as an open interface for realizing computing power external calling in the computing power service module, and the computing power template layer can be understood as abstracting diversified and complex computing power capability of the bottom layer and externally releasing a development execution environment template with unified computing power, wherein the development execution environment template comprises a computing power type, a computing power magnitude value and the like.

The initial computing power information may be understood as information related to the computing power provided for the service requirement, including computing power information of the processor, storage power information, and the like, which are not specifically limited herein.

In practical application, a computing power calling layer in the computing power service module provides computing power service for third-party application through a computing power API calling interface which is open to the outside; after receiving a calculation power calling request sent by a target user, a calculation power calling layer forwards initial calculation power information carried in the calculation power calling request to a calculation power template layer, so that the calculation power template layer can determine target calculation power information according to the initial calculation power information and a preset calculation power template, and further generate a target calculation power calling instruction for the target calculation power information, for example, the storage capacity of the initial calculation power information required to be called by the target user is 40G, and the templates with the storage capacity of the preset calculation power template are 20G, 50G and 100G, so that the calculation power template layer can select a template with the storage capacity of 50G from a plurality of calculation power templates as the calling requirement of the target calculation power, thereby ensuring that the required capacity is provided, the calculation resources are not wasted, and reasonable distribution can be realized; it should be noted that unified management is facilitated by presetting the computing power template to implement a management layer of the computing power network system architecture.

The computing network system provided by the embodiment of the specification realizes calling communication with an external third-party application through the computing calling layer, and efficiently and reliably creates computing service facing the third-party application with lower average computing operation cost and better matching with the computing service quality required by the application.

The computational power network system provided by the embodiment of the specification, wherein the computational power control module comprises a computational power measurement layer, a computational power strategy layer and a computational power addressing layer,

the computing power measurement layer is configured to receive the computing power routing tables uploaded on the computing power routing layers of the at least two computing power nodes and forward the computing power routing tables uploaded on the computing power routing layer of each computing power node to the computing power policy layer;

the computing power strategy layer is configured to search a computing power routing table uploaded by a computing power routing layer of each computing power node based on the received target computing power calling instruction, and determine a target computing power capability parameter and a target network capability parameter corresponding to the target computing power information in the computing power routing table uploaded by the computing power routing layer of each computing power node according to the target computing power information carried in the target computing power calling instruction;

the computational power strategy layer is further configured to determine a target computational power strategy based on the target computational power capability parameter and a target network capability parameter, and forward the target computational power strategy to the computational power addressing layer;

the computation power addressing layer is configured to determine a target computation power destination address and target routing path information from a preset routing address table based on the target computation power policy, wherein the preset routing address table comprises routing node marks and routing addresses of all computation power nodes.

Wherein, the computation power measuring layer can be understood as the computation power information of the collected computation power routing table from each computation power node; the calculation power strategy layer can be understood as a calculation power resource calling strategy which is comprehensively decided according to the calculation power information in the calculation power routing table of each calculation power node; the computation power addressing layer can be understood as inquiring a routing destination address corresponding to the target computation power node according to the target computation power node decided by the strategy layer.

In practical application, the computational force control module can be understood as a centralized intelligent computational force control center, has an intelligent center function, and completes overall computational force overall center control and intelligent scheduling; wherein, the computation force measuring layer in the computation force control module can collect the computation force routing table uploaded by the computation force routing layer in each computation force node, and forward the computation force routing table of each computation force node to the computation force strategy layer, after receiving the target computation force calling instruction sent by the computation force service module, the computation force strategy layer can search the computation force routing table uploaded by the computation force routing layer of each computation force node based on the target computation force calling instruction, and according to the current computation force state, i.e. computation capability parameter, network capability parameter, etc. in each current computation force node, decide the computation force node capable of providing the target computation force information, specify the specific target computation force strategy, and at the same time, forward the target computation force strategy to the computation force addressing layer in the computation force control module, which is convenient for realizing the addressing function subsequently, it needs to say that the specific decision mode can be the sum of weighting computation parameter and network parameter, the optimal execution node is solved to implement the calculation route forwarding of the service, and may also be implemented in various decision manners, and the specific decision manner is not specifically limited in the embodiments of the present specification.

Further, after receiving the currently decided target computation power strategy, a computation power addressing layer in the computation power control module can query a preset routing address table according to a target computation power destination address and target routing path information in the target computation power strategy to determine a routing address of a target computation power node and a specific routing path reaching the routing address; it should be noted that the calculation power policy in the calculation power control module may make a dynamic intelligent decision according to the calculation power routing table acquired in real time, and the specific dynamic decision manner in the embodiment of this specification is not limited at all.

In addition, the embodiments of the present specification also provide key information in the computation routing table, as shown in table 1 below, but not limited to the data types in the following table:

TABLE 1

It should be noted that, the same service corresponds to a unique service ID, and when different nodes in the network deploy the service, multiple copies of the service are distributed in the network, and multiple copies of the same service are mapped to the same service ID corresponding to multiple different service IP addresses. The service ID uses a service ID defined by the service layer (which may be a service ID translated in IP format) for compatibility with existing service layers.

In the computational power network system provided in the embodiment of the present specification, the computational power control module synthesizes the computational power and the routing capability through the obtained computational power routing table, and decides a better execution computational power node, so that not only can the path transmission delay and the computation delay be greatly reduced, but also the computation performance requirements of applications can be met.

To facilitate understanding of the comprehensive computation capability and routing capability of the computational force control module, and to decide a better execution computational force node to meet the service computational force usage requirement, reference may be made to fig. 2 specifically, and fig. 2 shows a schematic diagram of a service data packet forwarding flow of a computational force forwarding plane in a computational force network system provided in an embodiment of the present specification.

It should be noted that the computational power forwarding plane may be understood as a plurality of nodes that implement computational power provision, including an end node, an edge cloud node, a center cloud node, and the like; the service data packet may be understood as computing capability information (such as node information providing computing capability, computing capability information, computing capability type information, service identification and type information, etc.) provided for the user.

In practical application, a computation power policy layer in the computation power control layer can match with better computation resources (computation power capability parameters) and storage resources (network capability parameters) through a computation power request of a target service, and determine corresponding target computation cluster nodes (computation power execution nodes), that is, the computation power policy layer can search and match with a computation power routing table, select an optimal computation power table item in the computation power routing table, and obtain a corresponding computation power cluster ID supporting service computation, wherein the computation power request includes, but is not limited to, the computation resources, the storage resources and the like required by the execution service; after receiving the instruction for invoking the computing power, the plurality of nodes in the computing power forwarding plane can generate corresponding service data packets according to the computing power requirements carried in the computing power request, and forward the service data packets to the computing power control layer, so that the computing power requirements corresponding to services can be provided for users.

Further, in order to more effectively control the computation power information in each computation power node, the computation power network system provided in the embodiment of the present specification needs to create a computation power routing layer to create and generate a computation power routing table, so as to realize real-time understanding and mastering of the computation power and the network power of each computation power node; the at least two computation force nodes further comprise a computation force control sublayer and a computation force sensing layer,

the computing power sensing layer is configured to acquire computing power information of a local computing power node and a neighbor computing power node in real time and forward the computing power information to a computing power routing layer in the computing power node, wherein the computing power information comprises computing power capability parameters and network capability parameters;

the computing power routing layer is further configured to generate a computing power routing table corresponding to the local computing power node based on the computing power capability parameter and a network capability parameter, and forward the computing power routing table to the computing power control sublayer;

and the computation force control sublayer is configured to upload a computation force routing table corresponding to the local computation force node to a computation force measurement layer in the computation force control module.

The computing nodes can be understood as computing nodes in a distributed computing network architecture, such as segment computing nodes, edge cloud computing nodes and center cloud computing nodes; the computational power node in the embodiment of the present specification is not specifically limited, and any server device or server cluster that can provide a computational power demand may be regarded as a computational power node, and may also be understood as a computational power forwarding plane.

The computational force control sublayer can be understood as a distributed computational force control layer, can be a decentralized full-distributed computational force control layer, can also perform layered management and control, is deployed in computational force control functions of various distributed networks or intelligent edges of a universal terminal, and cooperates with a centralized computational force control layer to form an internal computational force system of the network.

The computing power perception layer can be understood as perception, measurement and application of computing power resource information of a local computing power node and a neighbor computing power node. The method realizes perception, measurement and application of computational power node resources of terminal equipment, network routing equipment and cloud computing clusters by constructing and deploying the computational power node equipment in a terminal equipment software layer, routing switching equipment, a center or edge data center cluster and the like. And the calculation force service layer and the calculation force control layer cooperate to realize calculation force distribution, addressing and scheduling of a single device or an area cluster.

In practical application, the calculation force sensing layer in each calculation force node can acquire calculation force information of a local calculation force node and a neighboring calculation force node in real time, and then forwards the calculation force information to the calculation force routing layer in the calculation force node, so that the calculation force routing layer can generate a calculation force routing table based on the calculation force capability parameters and the network capability parameters in the calculation force information, and the calculation force routing layer uploads the generated calculation force routing table to the calculation force control sublayer, that is, the calculation force control sublayer can control calculation force resources in the local calculation force node in real time, and resource distribution and control inside a resource cluster of the local calculation force node are facilitated. The computation force control sublayer uploads a computation force routing table of a local computation force node to a computation force measurement layer in the computation force control module, and then, through distributed computation force sensing, a creating and addressing function is provided for providing a fast on-demand computation force service for a large number of edge devices.

The computational force control sublayer can report the computational force routing table of each row examined by the node according to the computational force information so as to facilitate a subsequent computational force control module to make a target computational force strategy decision according to the reported computational force routing table; specifically, referring to fig. 3, fig. 3 is a schematic diagram illustrating a process of reporting computing power routing information of a computing power network system according to an embodiment of the present disclosure.

In practical application, the computational power forwarding plane can generate a computational power routing table from the computational power information of the node, including information such as application name, application type, computational capability, storage type, storage capability, computational power factor, connection capability and the like, and send the computational power routing table to the computational power control module; the computational power control module can also comprehensively decide computational power execution nodes according to the reported computational power routing tables of all computational power nodes, and then the computational power forwarding plane is responsible for responding to specific computational power calling instructions.

It should be noted that the computing power sensing layer senses the network state and the computing power position in real time based on the network layer, and the computing power network can combine with real-time information to realize rapid computing power discovery and routing no matter the traditional centralized cloud computing power or other computing power distributed in the network; the network comprehensively considers real-time network resource conditions and computing resource conditions based on SLA requirements of users, and quickly matches service flow to an optimal node through flexible and dynamic scheduling of the network, so that the network supports dynamic service to ensure user experience of services. When a data packet of a task is calculated, firstly, the calculation task type of the data packet is determined, and at least one other node corresponding to the calculation task type and the calculation performance corresponding to the other node are determined based on the corresponding relation of the calculation task type, the other calculation node and the calculation performance which is obtained in advance. The target node for execution is determined based on the computational performance of other nodes and network performance, such as link status, between the local node and other nodes. The address of the destination node, i.e. the routing destination address of the data packet, and then forwards the data packet based on the destination address.

In addition, the computation force routing layer is further configured to update the computation force routing table of the local computation force node according to the computation force information of the local computation force node and the adjacent computation force nodes acquired in real time.

In practical application, the computational power routing table in each computational power node can also record computational power resource information of neighbor nodes of the local computational power node, and in order to more accurately realize the scheduling of the whole network computational power resource, the computational power routing table of each computational power node also needs to be updated in real time, wherein the calculation and network information can be subjected to information propagation and synchronization by expanding the existing BGP, IGP protocols and the like, so that a centralized computational power control module can collect the updated computational power routing table to accurately realize the scheduling and control of the whole network computational power resource; it should be noted that, the manner of synchronization and update of the computing resource information between each computing node is not specifically limited in this embodiment.

Furthermore, the calculation routing layer can not only realize the updating of the self calculation routing table according to the local calculation nodes and the neighbor calculation nodes which are obtained in real time, but also realize the updating of the further calculation routing table depending on the grasp of the calculation control layer on the calculation capability of the global calculation node; specifically, referring to fig. 4, fig. 4 is a schematic flowchart illustrating a computational power route generation in a computational power network system according to an embodiment of the present disclosure.

In practical application, the computational force control layer receives the computational force information reported by all the computational force nodes and further has the capability of global grasping all the computational force nodes, so the computational force control layer can also send computational force routing information of the node and neighbor nodes associated with the node to each node, wherein the computational force routing information can comprise information such as a computational force node I D table, an application name, an application type, a computational type, computational capability, a storage type, a storage capability, a computational force factor, connection capability and the like; each computing node of the computation power forwarding plane can locally store the computation power routing information which is correspondingly issued, so as to update the content recorded in the local routing table.

The computing power network system provided by the embodiment of the specification realizes the recording and generation of computing power resources of the current computing power node through the computing power perception layer and the computing power routing layer in each computing power node, and is convenient for a subsequent control module to complete overall computing power overall center and intelligent scheduling according to the recorded computing power resources.

In addition, the computational power network system provided in the embodiments of the present specification further includes: a computing power orchestration management module, wherein the computing power orchestration module comprises a computing power modeling layer,

the calculation force modeling layer is configured to obtain initial calculation force information carried in the calculation force calling request, construct a preset calculation force template based on the initial calculation force information, and send the preset calculation force template to the calculation force template layer, wherein the preset calculation force template comprises preset calculation force capability parameter values and preset network capability parameter values.

The computing power arrangement management module can be understood to make a whole-network computing power control management strategy, realize computing power model construction, computing power authorization, computing power unified management and implementation deployment, need to perform unified management and arrangement aiming at computing power equipment and computing power routing of the whole network, and realize arrangement strategy interaction of a computing power arrangement layer and a computing power control layer through an API (application programming interface).

The computational power modeling layer can be understood as a process of realizing computational power template modeling according to the computational power resource calling requirement of the third-party application so as to realize the quantification of computational power resources.

In practical application, the calculation power modeling layer in the calculation power arrangement management module can also provide a preset calculation power template for the calculation power service module, wherein the preset calculation power template can be a template model constructed in advance according to calling requirements, each calculation power template can comprise preset calculation power capability parameter values and preset network capability parameter values, and the preset calculation power capability parameter values and the preset network capability parameter values can be understood as specific parameter values, so that the quantification of template parameters is realized, and the unified management and the provision are facilitated.

The computing power network system provided in the embodiment of the description constructs the preset computing power template through the computing power modeling layer in the computing power arrangement management module, so that a uniform computing power calling parameter value is determined through the preset computing power template in the following process, and management and control of the whole network computing power resource are facilitated.

Further, the computing power arrangement module also comprises an arrangement scheduling layer,

the scheduling layer is configured to acquire the target calculation power destination address and the target routing path information from the calculation power addressing layer of the calculation power control module based on a scheduling interface, and perform resource allocation on the target calculation power node corresponding to the target calculation power destination address based on the target routing path information.

The scheduling layer may be understood to perform scheduling on the global computing resources, for example, a scheduling target computing node provides corresponding computing resources.

In practical application, after determining a target calculation power destination address and target routing path information of a target calculation power node, an arrangement scheduling layer can perform resource calling and allocation on the target calculation power node through the target routing path information, under a centralized calculation power deployment scene, calculation power, storage and network resource and node information of an end, a side and a cloud are uniformly collected and distributed by a centralized orchestrator, and the centralized orchestrator combines a large-scale heterogeneous calculation power resource state and a network transmission state of a whole network according to application requirements, arranges an optimal forwarding and routing path and sends the optimal forwarding and routing path to calculation power forwarding nodes in the network.

Under a distributed computing power deployment scene, the capacity of intelligently arranging and scheduling heterogeneous computing power resources needs to be added to key computing power equipment in a network system. By the cooperation of distributed computational power nodes and a centralized computational power arranging and scheduling system, the large-scale heterogeneous computational power capability of the whole network is optimally arranged and scheduled, the method adapts to the complex future network environment, meets the aim of computational power scheduling and application as required, and really embodies the network computational power originality.

The computational power network system provided in the embodiment of the present specification utilizes the scheduling layer in the computational power scheduling management module to implement the specific scheduling and management of computational power nodes, and complete the computational power operation, the computational power service scheduling, and the management of computational power resources and network resources.

Further, the computing power arrangement module further comprises a life cycle management layer,

the life cycle management layer is configured to obtain the computing power routing table of each computing power node from the computing power control sublayer of each computing power node in real time, and record the resource state information of each computing power node based on the computing power routing table.

The lifecycle management layer may be understood to manage operations of the control system, including operations of system online, configuration, policy execution, soft/firmware update, and other system lifecycle management, such as recording the computing resource status of each computing node processor.

In practical application, the life cycle management layer in the computing power arrangement management layer can acquire the computing power routing table of each computing power node in real time from the computing power control sublayer of each computing power node, and further record the resource state information of each computing power node according to the computing power routing table of each computing power node, so that the management of the whole life cycle of each computing power node is realized.

In addition, the computing power network system provided by the embodiments of the present specification, the at least two computing power nodes further include a network protocol layer and a computing power resource layer,

and the computing force control sublayer of the target computing force node receives the computing force calling request based on the network protocol layer and calls the computing resources in the computing force resource layer based on the computing force calling request.

Wherein, the network protocol layer can be understood as a link layer/I P layer, and the computing resource layer can be understood as a computing resource, a storage resource, a network resource, etc. comprising the computing node.

In practical application, the scheduling layer of the computational power scheduling management layer can specifically realize scheduling of computational power resources of computational power nodes, after the computational power control layer determines called target computational power nodes, the scheduling layer can receive computational power calling requests to a computational power control sublayer in the target computational power nodes, can receive the calling requests according to a network protocol layer, and then completes the calling of the computational power resources based on the calling requests and the computation resources in the actual computational power resource layer.

In summary, the computational power network system provided in the embodiments of the present specification implements sensing, creating, addressing, modeling, and scheduling of computational power resources through the computational power control module, the computational power service module, the computational power scheduling module, and the computational power routing layer, the computational power sensing layer, and the computational power control sublayer in the computational power nodes, and implements dynamic sensing and on-demand scheduling of the computational power resources of the entire network through two deployment manners, i.e., centralized and distributed, to implement the state propagation and application scheduling of the computational power of the entire network.

Referring to fig. 5, fig. 5 is a flowchart illustrating a calculation power calling method according to an embodiment of the present disclosure, which includes the following steps.

It should be noted that the computing power calling method provided in the embodiments of the present specification may be applied to a computing power network system, where the computing power network system includes a computing power service module, a computing power control module, and at least two computing power nodes.

Step 502: the computing power service module generates a target computing power calling instruction based on computing power information carried in a computing power calling request sent by a target user.

In practical application, the computing power service module may further include a computing power template layer, which determines a quantitative target computing power resource value (including a computing parameter value and a network parameter value) from the computing power information carried in the computing power calling request, and then generates a target computing power calling instruction according to the target computing power resource value.

Step 504: and the calculation force control module acquires calculation force routing tables generated by the at least two calculation force nodes based on the target calculation force calling instruction, and determines a target calculation force strategy based on the calculation force routing table of each calculation force node.

And the target calculation force strategy comprises a target calculation force capability parameter and a target network capability parameter corresponding to the calculation force information.

It should be noted that, for the process of generating the calculation power routing table by each calculation power node in the calculation power invoking method provided in this embodiment and the process of determining the target calculation power policy according to the calculation power routing table, reference may be made to the description in the foregoing embodiment, and redundant description is not repeated here.

Step 506: and the calculation force control module determines a routing address of a target calculation force node from a preset routing address table based on the target calculation force strategy and calls a calculation resource corresponding to the calculation force information for the target user based on the routing address.

In practical application, after determining the routing address of the target computational power node, the computational power control module can send the routing address to a computational power scheduling layer in the computational power scheduling module to realize the calling of computational power resources of the target computational power node, wherein the computational power scheduling module can be understood as a module in a computational power network system and provides a part of computational power scheduling management for the whole computational power network system.

In the calculation power calling method provided in the embodiment of the present specification, the target calculation power node is determined through the decision of the calculation power control module on the target calculation power policy, and then the routing address of the target calculation power node is determined through the addressing function, so that the calculation power resource of the target calculation power node is called, and not only can the scheduling efficiency of the calculation power resource be improved, but also the delay of calling a routing path and the calculation delay are both greatly reduced, so as to meet the requirement of calling performance.

The computing power calling method provided in this specification is further described below with reference to fig. 6, taking an application of the computing power calling method in a computing power network system as an example. Fig. 6 is a schematic diagram illustrating a processing procedure of an algorithm call method according to an embodiment of the present disclosure.

Fig. 6 includes a computing power service module, a computing power arrangement management module, a computing power control module, and at least two computing power nodes (end, edge cloud, and center cloud), where the computing power service module includes multiple processing layers, and is respectively open for computing power capability and published for computing power template; the computing power arrangement management module comprises a plurality of processing layers which are respectively used for computing power modeling, arrangement scheduling and life cycle management; the calculation force control module comprises a plurality of processing layers which are respectively calculation force measurement, calculation force strategy and calculation force addressing; each calculation force node also comprises a plurality of processing layers which are respectively a distributed calculation force control layer, a calculation force routing layer, calculation force perception, a link layer/I P layer, calculation resources, storage resources and network resources; it should be noted that the computational force control module may interact with at least two computational force nodes, and each computational force node may also interact with each other, so as to update the computational force resource information of the adjacent node.

In practical application, a computing power open layer in a computing power service module can receive a computing power calling request of a third-party application through an interface, a computing power template issuing layer determines quantitative target computing power information based on a preset computing power template, generates a computing power calling instruction aiming at the target computing power information and sends the computing power calling instruction to a computing power control module, a computing power strategy layer in the computing power control module can obtain a computing power routing table generated by a computing power routing layer in the current full-network computing power node from a computing power measuring layer after receiving the computing power calling instruction, comprehensive analysis processing is carried out on computing power resource information of the computing power routing table in each computing power node, the target computing power node is intelligently decided, and it needs to be stated that information in the computing power routing table can be determined through computing power parameters and network power parameters recorded by a computing power sensing layer; the specific computational power node routing address of the target computational power node and the called routing path information can be inquired in a preset routing address table in the computational power addressing layer according to the specific node identification of the target computational power node; then sending the routing address of the computational power node and the called routing path information to an arrangement scheduling layer in the computational power arrangement management module to carry out computational power resource calling; and receiving the resource calling request through a link layer/I P layer in the target computing power node, and calling the computing resources, the storage resources, the network resources and the like in the target computing power node.

The distributed computational force control layer in each computational force node not only provides overall control of computational force resources for each computational force node, but also can be cooperatively operated with the computational force control module to realize end-to-end endogenous computational force control.

Corresponding to the above method embodiment, the present specification further provides an embodiment of an computing power invoking device, and fig. 7 shows a schematic structural diagram of the computing power invoking device provided in an embodiment of the present specification. As shown in fig. 7, the apparatus is applied to a computational power network system, which includes a computational power service module 702, a computational power control module 704 and at least two computational power nodes 706,

the computing power service module 702 is configured to generate a target computing power calling instruction based on computing power information carried in a computing power calling request sent by a target user;

the computational force control module 704 is configured to obtain computational force routing tables generated by the at least two computational force nodes 406 based on the target computational force calling instruction, and determine a target computational force policy based on the computational force routing table of each computational force node 706, where the target computational force policy includes a target computational force capability parameter and a target network capability parameter corresponding to the computational force information;

the computing force control module 704 is further configured to determine a routing address of a target computing force node from a preset routing address table based on the target computing force policy, and invoke a computing resource corresponding to the computing force information for the target user based on the routing address;

The computing power calling device provided in the embodiment of the present specification determines a target computing power node through a decision of a target computing power policy by a computing power control module, and further determines a routing address of the target computing power node through an addressing function, so as to realize calling of computing power resources of the target computing power node, which not only can improve scheduling efficiency of the computing power resources, but also can greatly reduce delay and computation delay of calling a routing path, so as to meet requirements of calling performance.

The above is an exemplary scheme of the computing power calling apparatus of the present embodiment. It should be noted that the technical solution of the calculation power calling device and the technical solution of the calculation power calling method belong to the same concept, and details of the technical solution of the calculation power calling device, which are not described in detail, can be referred to the description of the technical solution of the calculation power calling method.

FIG. 8 illustrates a block diagram of a computing device 800, according to one embodiment of the present description. The components of the computing device 800 include, but are not limited to, memory 810 and a processor 820. The processor 820 is coupled to the memory 810 via a bus 830, and the database 850 is used to store data.

Computing device 800 also includes access device 840, access device 840 enabling computing device 800 to communicate via one or more networks 860. Examples of such networks include the Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the internet. Access device 840 may include one or more of any type of network interface (e.g., a Network Interface Card (NIC)) whether wired or wireless, such as an ieee e802.11 Wireless Local Area Network (WLAN) wireless interface, a worldwide interoperability for microwave access (Wi-MAX) interface, an ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a bluetooth interface, a Near Field Communication (NFC) interface, and so forth.

In one embodiment of the present description, the above-described components of computing device 800, as well as other components not shown in FIG. 8, may also be connected to each other, such as by a bus. It should be understood that the block diagram of the computing device structure shown in FIG. 8 is for purposes of example only and is not limiting as to the scope of the description. Those skilled in the art may add or replace other components as desired.

Computing device 800 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), a mobile phone (e.g., smartphone), a wearable computing device (e.g., smartwatch, smartglasses, etc.), or other type of mobile device, or a stationary computing device such as a desktop computer or PC. Computing device 800 may also be a mobile or stationary server.

Wherein the processor 820 is configured to execute computer-executable instructions that, when executed by the processor, implement the steps of the computational power calling method described above.

The above is an illustrative scheme of a computing device of the present embodiment. It should be noted that the technical solution of the computing device and the technical solution of the computing power calling method belong to the same concept, and details that are not described in detail in the technical solution of the computing device can be referred to the description of the technical solution of the computing power calling method.

An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which when executed by a processor implement the steps of the computational calling method described above.

The above is an illustrative scheme of a computer-readable storage medium of the present embodiment. It should be noted that the technical solution of the storage medium belongs to the same concept as the technical solution of the computing power calling method, and details that are not described in detail in the technical solution of the storage medium can be referred to the description of the technical solution of the computing power calling method.

An embodiment of the present specification further provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the computational power calling method.

The above is an illustrative scheme of a computer program of the present embodiment. It should be noted that the technical solution of the computer program and the technical solution of the computing power calling method belong to the same concept, and details that are not described in detail in the technical solution of the computer program can be referred to the description of the technical solution of the computing power calling method.

The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The computer instructions comprise computer program code which may be in the form of source code, object code, an executable file or some intermediate form, or the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.

It should be noted that, for the sake of simplicity, the foregoing method embodiments are described as a series of acts, but those skilled in the art should understand that the present embodiment is not limited by the described acts, because some steps may be performed in other sequences or simultaneously according to the present embodiment. Further, those skilled in the art should also appreciate that the embodiments described in this specification are preferred embodiments and that acts and modules referred to are not necessarily required for an embodiment of the specification.

In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

The preferred embodiments of the present specification disclosed above are intended only to aid in the description of the specification. Alternative embodiments are not exhaustive and do not limit the invention to the precise embodiments described. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the embodiments and the practical application, to thereby enable others skilled in the art to best understand and utilize the embodiments. The specification is limited only by the claims and their full scope and equivalents.

Claims

1. A computing force network system comprising: the system comprises a calculation force service module, a calculation force control module and at least two calculation force nodes, wherein the at least two calculation force nodes comprise a calculation force routing layer;

2. The computing power network system of claim 1, the computing power service module comprising a computing power calling layer, a computing power template layer,

3. The computing network system of claim 1, the computing force control module comprising a computing force measurement layer, a computing force policy layer, and a computing force addressing layer,

4. The computational network system of claim 3, the at least two computational nodes further comprising a computational control sublayer, a computational sense layer,

5. The computing power network system of claim 4, the computing power routing layer further configured to update the computing power routing table of the local computing power node according to the computing power information of the local computing power node and the neighboring computing power nodes acquired in real time.

6. The computing power network system of claim 2, further comprising: a computing power orchestration management module, wherein the computing power orchestration module comprises a computing power modeling layer,

7. The computing power network system of claim 3, the computing power orchestration module further comprising an orchestration scheduling layer,

8. The computing power network system of claim 6 or 7, the computing power orchestration module further comprising a lifecycle management layer,

the life cycle management layer is configured to obtain a calculation force routing table of each calculation force node in real time from a calculation force control sublayer of each calculation force node, and record resource state information of each calculation force node based on the calculation force routing table.

9. The computing power network system of claim 7, the at least two computing power nodes further comprising a network protocol layer, a computing power resource layer,

10. A computing power calling method is applied to a computing power network system, the computing power network system comprises a computing power service module, a computing power control module and at least two computing power nodes,

the calculation force service module generates a target calculation force calling instruction based on calculation force information carried in a calculation force calling request sent by a target user;

11. A computing power calling device is applied to a computing power network system, the computing power network system comprises a computing power service module, a computing power control module and at least two computing power nodes,

12. A computing device, comprising:

a memory and a processor;

the memory is configured to store computer-executable instructions and the processor is configured to execute the computer-executable instructions, which when executed by the processor, implement the steps of the computer-implemented method of claim 10.

13. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, perform the steps of the computer-readable call method of claim 10.