CN114980250A - Computing power routing system and method based on SRv6 - Google Patents

Abstract

The invention provides a calculation routing system and method based on SRv6, wherein the system comprises: the system comprises a calculation routing controller, a calculation access gateway, a relay routing node and a calculation service gateway; the computing power routing controller is used for determining a computing power service SRv6SID and a forwarding action strategy of the computing power service gateway based on the computing power service information and the computing power routing information; the computing power access gateway is used for forwarding computing power service flow based on an anycast IPv6 address and a forwarding action strategy of target computing power service carried by the computing power service flow; the relay routing node is used for forwarding the computing power service flow based on the forwarding action strategy; and the computing service gateway is used for forwarding computing service traffic based on the forwarding action strategy and the computing service SRv6 SID. The calculation power routing system and method based on SRv6 provided by the invention can realize the efficient calling of calculation power network resources on the premise of slightly changing the infrastructure of the calculation power network.

Description

Translated fromChinese

基于SRv6的算力路由系统及方法Computing power routing system and method based on SRv6

技术领域technical field

本发明涉及计算机技术领域，尤其涉及一种基于SRv6的算力路由系统及方法。The present invention relates to the field of computer technology, and in particular, to an SRv6-based computing power routing system and method.

背景技术Background technique

随着移动通信技术的发展，移动通信网络(例如5G网络)已经具备了大带宽低时延的空口性能特性。但传统的基于中心云的计算模式，仍需要将业务数据传输到远端服务器进行处理，带来大量的传输时延，难以为用户提供良好的服务质量保障，也难以发挥移动通信网络性能提升的优势。With the development of mobile communication technologies, mobile communication networks (such as 5G networks) already have air interface performance characteristics of large bandwidth and low latency. However, the traditional cloud-based computing model still needs to transmit the business data to the remote server for processing, which brings a lot of transmission delay. Advantage.

分布式计算与边缘计算可以在一定程度上利用移动通信网络性能提升的优势，但随着泛在计算节点的部署，如何高效调用协同这些算力资源和网络资源成为亟待解决的重要技术问题。Distributed computing and edge computing can take advantage of the performance improvement of mobile communication networks to a certain extent, but with the deployment of ubiquitous computing nodes, how to efficiently call and coordinate these computing resources and network resources has become an important technical problem that needs to be solved urgently.

目前，可以通过算力网络解决泛在算力和网络的协同调度问题。算力网络技术中，算力路由方法是算力业务流量调度的关键技术，可以决定能否充分利用泛在部署的算力网络资源。算力路由方法的本质是通过网络路由层面的调度控制实现对已部署算力服务的高效调用，At present, the collaborative scheduling problem of ubiquitous computing power and the network can be solved through the computing power network. In the computing power network technology, the computing power routing method is the key technology of computing power business traffic scheduling, which can determine whether the ubiquitous deployment of computing power network resources can be fully utilized. The essence of the computing power routing method is to achieve efficient invocation of the deployed computing power services through scheduling control at the network routing level.

但现有算力路由方法的实施，通常需要对算力网络的基础设施进行大量更新升级，难以实现向算力网络的平滑演进。However, the implementation of the existing computing power routing method usually requires a large number of upgrades to the infrastructure of the computing power network, and it is difficult to achieve a smooth evolution to the computing power network.

发明内容SUMMARY OF THE INVENTION

本发明提供一种基于SRv6的算力路由系统及方法，用以解决现有技术中需要对算力网络的基础设施进行大量更新升级的缺陷，实现在对算力网络的基础设施较小改动的前提下，实现对算力服务的高效调用。The present invention provides a computing power routing system and method based on SRv6, which is used to solve the defect in the prior art that the infrastructure of the computing power network needs to be updated and upgraded in large quantities, and realizes the need for minor changes to the infrastructure of the computing power network. On the premise, to achieve efficient invocation of computing power services.

本发明提供一种基于SRv6的算力路由系统，包括：算力路由控制器、至少一个算力接入网关、至少一个中继路由节点和至少一个算力服务网关；The present invention provides a computing power routing system based on SRv6, comprising: a computing power routing controller, at least one computing power access gateway, at least one relay routing node and at least one computing power service gateway;

所述算力路由控制器，用于基于算力服务信息和算力路由信息，确定各所述算力服务网关的算力服务SRv6 SID、多条算力路由路径及各所述算力接入网关的第一转发动作策略和各所述算力服务网关的第二转发动作策略；向每一算力接入网关下发所述每一算力接入网关的第一转发动作策略；向每一算力服务网关下发所述每一算力服务网关的算力服务SRv6 SID和第二转发动作策略；The computing power routing controller is configured to determine, based on the computing power service information and the computing power routing information, the computing power service SRv6 SID of each of the computing power service gateways, multiple computing power routing paths, and each computing power access the first forwarding action policy of the gateway and the second forwarding action policy of each of the computing power service gateways; delivering the first forwarding action policy of each computing power access gateway to each computing power access gateway; A computing power service gateway delivers the computing power service SRv6 SID and the second forwarding action policy of each computing power service gateway;

所述算力接入网关，用于基于用户设备发送的算力业务流量携带的目标算力服务的任播IPv6地址和所述算力接入网关的所述第一转发动作策略，将所述算力业务流量封装对应的SR扩展头之后，将所述算力业务流量转发至相应的中继路由节点；The computing power access gateway is configured to, based on the anycast IPv6 address of the target computing power service carried by the computing power service traffic sent by the user equipment and the first forwarding action policy of the computing power access gateway, convert the computing power access gateway. After the computing power service traffic is encapsulated with the corresponding SR extension header, the computing power service traffic is forwarded to the corresponding relay routing node;

所述中继路由节点，用于基于所述中继路由节点的第三转发动作策略和所述对应的SR扩展头，将接收的封装所述对应的SR扩展头的所述算力业务流量进行转发；The relay routing node is configured to, based on the third forwarding action policy of the relay routing node and the corresponding SR extension header, perform the received computing power service traffic encapsulating the corresponding SR extension header. Forward;

所述算力服务网关，用于基于所述算力服务网关的第二转发动作策略、所述对应的SR扩展头和所述算力服务网关的算力服务SRv6 SID，将所述算力业务流量转发至目标算力服务实例节点；The computing power service gateway is configured to convert the computing power service based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header, and the computing power service SRv6 SID of the computing power service gateway. The traffic is forwarded to the target computing service instance node;

其中，所述算力服务信息，包括各算力服务实例节点的负载状态信息、服务能力信息、提供的算力服务的算力服务ID和任播IP地址及连接的所述算力服务网关的IP地址；所述算力路由信息，包括各所述算力接入网关、各所述中继路由节点和各所述算力服务网关的拓扑信息、链路带宽信息、链路时延信息及链路丢包率信息；所述算力路由路径，指从一个所述算力接入网关，经至少一个所述中继路由节点到一个所述算力服务网关的路径；所述目标算力服务的任播IPv6地址，是基于网络服务提供商分配给所述目标算力服务的算力服务提供商的第一信息、所述目标算力服务的ID和补充参数确定的；所述算力服务网关的算力服务SRv6 SID，是基于所述网络服务提供商分配给所述算力服务网关的第二信息、与所述算力服务网关通信连接的各算力服务实例节点提供的算力服务的ID和补充参数确定的；所述目标算力服务实例节点，包括至少一个与所述算力服务网关通信连接的算力服务实例节点。The computing power service information includes the load status information, service capability information of each computing power service instance node, the computing power service ID and anycast IP address of the computing power service provided, and the connected computing power service gateway. IP address; the computing power routing information, including the topology information, link bandwidth information, link delay information and Link packet loss rate information; the computing power routing path refers to the path from one of the computing power access gateways to one of the computing power service gateways via at least one of the relay routing nodes; the target computing power The anycast IPv6 address of the service is determined based on the first information of the computing power service provider assigned by the network service provider to the target computing power service, the ID of the target computing power service, and supplementary parameters; the computing power The computing power service SRv6 SID of the service gateway is based on the second information allocated to the computing power service gateway by the network service provider, and the computing power provided by each computing power service instance node communicatively connected to the computing power service gateway. The ID of the service and the supplementary parameters are determined; the target computing power service instance node includes at least one computing power service instance node that is communicatively connected to the computing power service gateway.

根据本发明提供的一种基于SRv6的算力路由系统，所述算力路由控制器，还用于基于所述算力服务信息和所述算力路由信息，确定每一条所述算力路由路径的优先级和/或权重；According to a computing power routing system based on SRv6 provided by the present invention, the computing power routing controller is further configured to determine each of the computing power routing paths based on the computing power service information and the computing power routing information priority and/or weight;

所述算力接入网关的所述第一转发动作策略，包括经过所述算力接入网关的所述算力路由路径的优先级和/或权重。The first forwarding action policy of the computing power access gateway includes the priority and/or weight of the computing power routing path passing through the computing power access gateway.

根据本发明提供的一种基于SRv6的算力路由系统，所述算力路由控制器，还用于基于新的算力服务信息和新的算力路由信息，确定上一次下发的各第一转发动作策略和各第二转发动作策略是否满足需求；在不满足需求的情况下，基于所述新的算力服务信息和所述新的算力路由信息，重新确定多条算力路由路径及各算力接入网关的第一转发动作策略和各算力服务网关的转发动作策略；向每一算力接入网关下发重新确定的所述每一算力接入网关的转发动作策略；向每一算力服务网关下发重新确定的所述每一算力服务网关的第二转发动作策略。According to a computing power routing system based on SRv6 provided by the present invention, the computing power routing controller is further configured to determine each first issued last time based on new computing power service information and new computing power routing information. Whether the forwarding action strategy and each second forwarding action strategy meet the requirements; if the requirements are not met, based on the new computing power service information and the new computing power routing information, re-determine multiple computing power routing paths and the first forwarding action policy of each computing power access gateway and the forwarding action policy of each computing power service gateway; delivering the re-determined forwarding action policy of each computing power access gateway to each computing power access gateway; The re-determined second forwarding action policy of each computing power serving gateway is delivered to each computing power serving gateway.

根据本发明提供的一种基于SRv6的算力路由系统，所述算力路由控制器，还用于基于算力服务信息和算力路由信息，确定各所述中继路由节点的第三转发动作策略；向每一中继路由节点下发所述每一中继路由节点的第三转发动作策略。According to a computing power routing system based on SRv6 provided by the present invention, the computing power routing controller is further configured to determine the third forwarding action of each relay routing node based on the computing power service information and the computing power routing information strategy; delivering the third forwarding action strategy of each relay routing node to each relay routing node.

根据本发明提供的一种基于SRv6的算力路由系统，所述算力路由控制器，还用于从算力服务编排器，获取所述算力服务信息。According to a computing power routing system based on SRv6 provided by the present invention, the computing power routing controller is further configured to obtain the computing power service information from the computing power service orchestrator.

根据本发明提供的一种基于SRv6的算力路由系统，各所述算力接入网关、各所述中继路由节点和各所述算力服务网关，还用于维护链路状态测量会话，以获取所述链路时延信息及链路丢包率信息；According to a computing power routing system based on SRv6 provided by the present invention, each of the computing power access gateways, each of the relay routing nodes and each of the computing power service gateways is further configured to maintain a link state measurement session, to obtain the link delay information and link packet loss rate information;

所述算力路由控制器，还用于从各所述算力接入网关、各所述中继路由节点和各所述算力服务网关，获取所述算力路由信息。The computing power routing controller is further configured to acquire the computing power routing information from each of the computing power access gateways, each of the relay routing nodes, and each of the computing power service gateways.

根据本发明提供的一种基于SRv6的算力路由系统，所述算力接入网关，还用于基于获取到的所述链路时延信息及链路丢包率信息，更新所述算力接入网关的所述第一转发动作策略中的所述算力路由路径的优先级和/或权重。According to a computing power routing system based on SRv6 provided by the present invention, the computing power access gateway is further configured to update the computing power based on the acquired link delay information and link packet loss rate information The priority and/or weight of the computing power routing path in the first forwarding action policy of the access gateway.

根据本发明提供的一种基于SRv6的算力路由系统，所述链路状态测量会话基于的协议包括Internet控制报文协议或双向主动测量协议。According to an SRv6-based computing power routing system provided by the present invention, the protocol based on the link state measurement session includes the Internet Control Message Protocol or the Bidirectional Active Measurement Protocol.

本发明还提供一种基于上述任一种所述基于SRv6的算力路由系统的基于SRv6的算力路由方法，包括：包括：The present invention also provides an SRv6-based computing power routing method based on any of the above-mentioned SRv6-based computing power routing systems, including: including:

在调用单个目标算力服务的情况下，算力路由控制器基于算力服务信息和算力路由信息，确定各算力服务网关的算力服务SRv6 SID、多条算力路由路径及各算力接入网关的第一转发动作策略和各所述算力服务网关的第二转发动作策略；向每一算力接入网关下发所述每一算力接入网关的第一转发动作策略；向每一算力服务网关下发所述每一算力服务网关的算力服务SRv6 SID和第二转发动作策略；In the case of invoking a single target computing power service, the computing power routing controller determines the computing power service SRv6 SID, multiple computing power routing paths and each computing power of each computing power service gateway based on the computing power service information and computing power routing information. a first forwarding action policy of the access gateway and a second forwarding action policy of each of the computing power service gateways; delivering the first forwarding action policy of each computing power access gateway to each computing power access gateway; delivering the computing power service SRv6 SID and the second forwarding action policy of each computing power service gateway to each computing power service gateway;

所述算力接入网关基于用户设备发送的算力业务流量携带的目标算力服务的任播IPv6地址和所述算力接入网关的所述第一转发动作策略，将所述算力业务流量封装对应的SR扩展头之后，将封装所述对应的SR扩展头的所述算力业务流量转发至相应的中继路由节点；The computing power access gateway transfers the computing power service based on the anycast IPv6 address of the target computing power service carried by the computing power service traffic sent by the user equipment and the first forwarding action policy of the computing power access gateway. After the traffic encapsulates the corresponding SR extension header, forward the computing power service traffic encapsulated with the corresponding SR extension header to the corresponding relay routing node;

中继路由节点基于所述中继路由节点的第三转发动作策略和所述对应的SR扩展头，将接收的封装所述对应的SR扩展头的所述算力业务流量进行转发；The relay routing node forwards the received computing power service traffic encapsulating the corresponding SR extension header based on the third forwarding action policy of the relay routing node and the corresponding SR extension header;

所述算力服务网关基于所述算力服务网关的第二转发动作策略、所述对应的SR扩展头和所述算力服务网关的算力服务SRv6 SID，将所述算力业务流量转发至目标算力服务实例节点。The computing power service gateway forwards the computing power service traffic to the computing power service gateway based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header and the computing power service The target computing service instance node.

根据本发明提供的一种基于SRv6的算力路由方法，还包括：A computing power routing method based on SRv6 provided by the present invention, further comprising:

在调用包括多个所述目标算力服务的算力服务功能链的情况下，算力路由控制器基于所述算力服务信息、所述算力路由信息和所述目标算力服务的调用顺序，确定各算力服务网关的算力服务SRv6 SID、多条算力路由路径及各算力接入网关的第一转发动作策略和各所述算力服务网关的第二转发动作策略；向每一算力接入网关下发所述每一算力接入网关的第一转发动作策略；向每一算力服务网关下发所述每一算力服务网关的算力服务SRv6 SID和第二转发动作策略；In the case of invoking a computing power service function chain including a plurality of the target computing power services, the computing power routing controller is based on the computing power service information, the computing power routing information and the calling sequence of the target computing power service. , determine the computing power service SRv6 SID of each computing power service gateway, multiple computing power routing paths, and the first forwarding action strategy of each computing power access gateway and the second forwarding action strategy of each computing power service gateway; A computing power access gateway delivers the first forwarding action policy of each computing power access gateway; delivers the computing power service SRv6 SID and the second computing power service gateway of each computing power service gateway to each computing power service gateway forwarding action strategy;

所述算力接入网关基于接收的算力业务流量携带的第一个目标算力服务的任播IPv6地址和所述算力接入网关的所述第一转发动作策略，将所述算力业务流量封装对应的SR扩展头之后，将封装所述对应的SR扩展头的所述算力业务流量转发至相应的中继路由节点；The computing power access gateway transfers the computing power based on the anycast IPv6 address of the first target computing power service carried by the received computing power service traffic and the first forwarding action policy of the computing power access gateway. After the service traffic encapsulates the corresponding SR extension header, forward the computing power service traffic encapsulated with the corresponding SR extension header to the corresponding relay routing node;

所述中继路由节点基于所述中继路由节点的第三转发动作策略和所述对应的SR扩展头，将接收的封装所述对应的SR扩展头的所述算力业务流量进行转发；The relay routing node forwards the received computing power service traffic encapsulating the corresponding SR extension header based on the third forwarding action policy of the relay routing node and the corresponding SR extension header;

所述算力服务网关基于所述算力服务网关的第二转发动作策略、所述对应的SR扩展头和所述算力服务网关的算力服务SRv6 SID，将所述算力业务流量转发至目标算力服务实例节点；The computing power service gateway forwards the computing power service traffic to the computing power service gateway based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header and the computing power service The target computing service instance node;

所述目标算力服务实例节点更新所述算力业务流量，并更新所述算力业务流量的目的地址为下一所述目标算力服务的任播IPv6地址；The target computing power service instance node updates the computing power service flow, and updates the destination address of the computing power service flow to be the anycast IPv6 address of the next target computing power service;

所述算力服务网关对所述目标算力服务实例节点返回的更新后的所述算力业务流量封装所述对应的SR扩展头，基于所述算力服务网关的第二转发动作策略和所述对应的SR扩展头，对封装所述对应的SR扩展头的所述算力业务流量进行转发。The computing power service gateway encapsulates the corresponding SR extension header for the updated computing power service flow returned by the target computing power service instance node, based on the second forwarding action policy of the computing power service gateway and the The corresponding SR extension header is forwarded, and the computing power service traffic encapsulated with the corresponding SR extension header is forwarded.

本发明提供的基于SRv6的算力路由系统及方法，基于算力服务的任播IP地址和算力服务网关的算力服务SRv6 SID进行算力业务流量的转发，对算力网络的基础设施的改动更小，能够在对算力网络的基础设施较小改动的前提下，实现算力网络资源更高效的调用，能通过算力路由控制器实现了对算力服务的高效调用，即“以网取算”。The SRv6-based computing power routing system and method provided by the present invention, based on the anycast IP address of the computing power service and the computing power service SRv6 SID of the computing power service gateway. The changes are smaller, and the computing power network resources can be called more efficiently under the premise of minor changes to the computing power network infrastructure. network calculation".

附图说明Description of drawings

为了更清楚地说明本发明或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。In order to explain the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are of the present invention. For some embodiments of the present invention, for those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort.

图1是本发明提供的基于SRv6的算力路由系统的结构示意图；1 is a schematic structural diagram of an SRv6-based computing power routing system provided by the present invention;

图2是本发明提供的基于SRv6的算力路由方法的流程示意图之一；2 is one of the schematic flowcharts of the SRv6-based computing power routing method provided by the present invention;

图3是本发明提供的基于SRv6的算力路由方法的流程示意图之二。FIG. 3 is the second schematic flowchart of the SRv6-based computing power routing method provided by the present invention.

具体实施方式Detailed ways

为使本发明的目的、技术方案和优点更加清楚，下面将结合本发明中的附图，对本发明中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。In order to make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention. , not all examples. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

在本发明实施例的描述中，术语“第一”、“第二”、“第三”仅用于描述目的，而不能理解为指示或暗示相对重要性，且不涉及顺序。In the description of the embodiments of the present invention, the terms "first", "second" and "third" are only used for the purpose of description, and should not be construed as indicating or implying relative importance, and do not involve sequence.

下面结合图1至图3描述本发明提供的基于SRv6的算力路由系统及方法。The SRv6-based computing power routing system and method provided by the present invention will be described below with reference to FIG. 1 to FIG. 3 .

图1是本发明提供的基于SRv6的算力路由系统的结构示意图。如图1所示，该系统可以包括：算力路由控制器101、至少一个算力接入网关102、至少一个中继路由节点103和至少一个算力服务网关104。FIG. 1 is a schematic structural diagram of an SRv6-based computing power routing system provided by the present invention. As shown in FIG. 1 , the system may include: a computing power routing controller 101 , at least one computingpower access gateway 102 , at least onerelay routing node 103 , and at least one computingpower service gateway 104 .

具体地，算力路由控制器101、算力接入网关102、中继路由节点103和算力服务网关104均基于IPv6转发平面的段路由(Segment Routing over IPv6，SRv6)工作Specifically, the computing power routing controller 101, the computingpower access gateway 102, therelay routing node 103, and the computingpower service gateway 104 all work based on the segment routing (Segment Routing over IPv6, SRv6) of the IPv6 forwarding plane

可选地，算力路由控制器101的数量可以为一个。算力路由控制器101，可以分别与每一算力接入网关102通信连接。算力路由控制器101，可以分别与每一算力服务网关104通信连接。Optionally, the number of computing power routing controllers 101 may be one. The computing power routing controller 101 may be in communication connection with each computingpower access gateway 102 respectively. The computing power routing controller 101 may be in communication connection with each computingpower service gateway 104 respectively.

可选地，算力路由控制器101，可以分别与每一中继路由节点103通信连接。算力路由控制器101与中继路由节点103之间可以通过常用的接口进行数据传输。Optionally, the computing power routing controller 101 may be in communication connection with eachrelay routing node 103 respectively. Data transmission can be performed between the computing power routing controller 101 and therelay routing node 103 through a common interface.

可选地，算力路由控制器101，可以与用于各控制中继路由节点103的SDN(软件定义网络，Software Defined Network)路由控制器通信连接。Optionally, the computing power routing controller 101 may be connected in communication with an SDN (Software Defined Network, Software Defined Network) routing controller for controlling eachrelay routing node 103 .

可选地，算力路由控制器101的数量可以为多个。每一算力路由控制器101，可以至少与一个其他算力路由控制器101通信连接。每一算力接入网关102，可以至少与一个算力路由控制器101通信连接。每一中继路由节点103，可以至少与一个算力路由控制器101通信连接。每一算力服务网关104，可以至少与一个算力路由控制器101通信连接。SDN路由控制器，可以至少与一个算力路由控制器101通信连接。每一算力服务网关104，可以不与其他算力服务网关104通信连接，也可以与至少一个其他算力服务网关104通信连接。Optionally, the number of computing power routing controllers 101 may be multiple. Each computing power routing controller 101 may be connected in communication with at least one other computing power routing controller 101 . Each computingpower access gateway 102 may be in communication connection with at least one computing power routing controller 101 . Eachrelay routing node 103 may be in communication connection with at least one computing power routing controller 101 . Each computingpower service gateway 104 may be in communication connection with at least one computing power routing controller 101 . The SDN routing controller may be connected in communication with at least one computing power routing controller 101 . Each computingpower service gateway 104 may not be in communication connection with other computingpower service gateways 104 , or may be in communication connection with at least one other computingpower service gateway 104 .

可选地，算力路由控制器101与算力接入网关102之间可以通过常用的接口进行数据传输。Optionally, data transmission may be performed between the computing power routing controller 101 and the computingpower access gateway 102 through a common interface.

可选地，算力路由控制器101与中继路由节点103之间可以通过常用的接口进行数据传输。Optionally, data transmission may be performed between the computing power routing controller 101 and therelay routing node 103 through a common interface.

可选地，算力路由控制器101与算力服务网关104之间可以通过常用的接口进行数据传输。Optionally, data transmission may be performed between the computing power routing controller 101 and the computingpower service gateway 104 through a common interface.

可选地，算力路由控制器101与SDN路由控制器之间可以通过常用的接口进行数据传输。Optionally, data transmission may be performed between the computing power routing controller 101 and the SDN routing controller through a common interface.

可选地，算力路由控制器101与SDN路由控制器之间可以通过常用的接口进行数据传输。Optionally, data transmission may be performed between the computing power routing controller 101 and the SDN routing controller through a common interface.

每一算力接入网关102，可以至少与一个中继路由节点103通信连接；每一中继路由节点103，可以至少与一个其他中继路由节点103通信连接，从而构成云间网络；每一算力服务网关104，可以至少与一个中继路由节点103通信连接。Each computingpower access gateway 102 can be in communication connection with at least onerelay routing node 103; eachrelay routing node 103 can be in communication connection with at least one otherrelay routing node 103 to form an inter-cloud network; eachrelay routing node 103 can be in communication connection with at least one otherrelay routing node 103; The computingpower service gateway 104 may be in communication connection with at least onerelay routing node 103 .

可以理解的是，图1示例性地示出了该算力路由系统包括的各设备之间的连接关系，但该算力路由系统包括的各设备之间的连接关系不限于图1所示出的情形。It can be understood that FIG. 1 exemplarily shows the connection relationship between the devices included in the computing power routing system, but the connection relationship between the devices included in the computing power routing system is not limited to that shown in FIG. 1 . situation.

可选地，算力服务网关104、算力接入网关101或中继路由节点102可以属于逻辑设备也可以是物理设备。算力服务网关104、算力接入网关101或中继路由节点102可以包括但不限于路由器设备、用户驻地设备(Customer premises equipment，CPE)、用户面功能(User Plane Function，UPF)设备或宽带接入服务器(Broadband Remote Access Server，BRAS)等各类物理设备承载其中一种或多种逻辑设备功能。Optionally, the computingpower service gateway 104, the computing power access gateway 101 or therelay routing node 102 may belong to a logical device or a physical device. The computingpower service gateway 104, the computing power access gateway 101 or therelay routing node 102 may include but are not limited to router equipment, customer premises equipment (CPE), user plane function (User Plane Function, UPF) equipment or broadband Various physical devices such as an access server (Broadband Remote Access Server, BRAS) carry one or more of the functions of the logical device.

SRv6，简单来讲即分段路由(Segment Routing，SR)+IPv6，是新一代IP承载协议。SRv6采用IPv6转发技术，通过灵活的IPv6扩展头，实现网络可编程。SRv6简化了网络协议类型，具有良好的扩展性和可编程性，可满足更多新业务的多样化需求，提供高可靠性，在云业务中有良好的应用前景。SRv6, simply referred to as Segment Routing (SR) + IPv6, is a new generation of IP bearer protocols. SRv6 adopts IPv6 forwarding technology and realizes network programmability through flexible IPv6 extension headers. SRv6 simplifies the types of network protocols, has good scalability and programmability, can meet the diverse needs of more new services, provides high reliability, and has good application prospects in cloud services.

SR技术的核心思想是将报文转发路径切割为不同的分段，并在路径起始点往报文中插入分段信息，中间节点只需要按照报文里携带的分段信息转发即可。这样的路径分段，称之为“Segment”，并通过SID(Segment Identifier，段标识)来标识。可以将多个Segment组合起来，得到Segment序列，从而形成SRv6路径，即路径可编辑。The core idea of SR technology is to cut the packet forwarding path into different segments, and insert segment information into the packet at the starting point of the path. Intermediate nodes only need to forward the packet according to the segment information carried in the packet. Such path segments are called "Segment" and are identified by SID (Segment Identifier, segment identifier). Multiple segments can be combined to obtain segment sequences to form SRv6 paths, that is, paths can be edited.

为基于IPv6转发平面实现SR技术，在IPv6路由扩展头新增SRH(Segment RoutingHeader，SR扩展头)。该扩展头指定一个IPv6的显式路径，存储IPv6的Segment List(段列表)信息。Segment List即对段和网络节点进行有序排列得到的一条转发路径。报文转发时，依靠Segments Left(下一跳的分段)和Segment List字段共同决定IPv6目的地址(IPv6DA)信息，从而指导报文的转发路径和行为。Segment序列，可以用Segment List表示。To implement the SR technology based on the IPv6 forwarding plane, an SRH (Segment Routing Header, SR extension header) is added to the IPv6 routing extension header. This extension header specifies an explicit path for IPv6 and stores IPv6 Segment List (segment list) information. The Segment List is a forwarding path obtained by orderly arranging segments and network nodes. When a packet is forwarded, the IPv6 destination address (IPv6DA) information is jointly determined by the Segments Left (segment of the next hop) and the Segment List field, so as to guide the forwarding path and behavior of the packet. Segment sequence, which can be represented by Segment List.

SRv6 Segment定义了SRv6网络编程中的网络指令，指示报文要去哪以及怎么去。标识SRv6 Segment的ID被称为SRv6 SID，SRv6 SID即前述SID。SRv6 SID是一个128bit的值，为IPv6地址形式，可以由Locator、Function和Arguments三部分组成。The SRv6 Segment defines the network instructions in SRv6 network programming, indicating where and how the packets are going. The ID that identifies the SRv6 Segment is called the SRv6 SID, and the SRv6 SID is the aforementioned SID. The SRv6 SID is a 128-bit value in the form of an IPv6 address, which can be composed of Locator, Function, and Arguments.

Locator字段：具有定位功能，用于提供IPv6的路由能力。报文通过Locator字段实现寻址转发。此外，Locator字段对应的路由也是可聚合的。Locator field: It has the positioning function and is used to provide IPv6 routing capability. Packets are forwarded by addressing through the Locator field. In addition, the route corresponding to the Locator field is also aggregable.

Function字段：用于表达该设备指令要执行的转发动作。不同的转发行为由Function字段的不同的值来表达。Function field: used to express the forwarding action to be performed by the device command. Different forwarding behaviors are expressed by different values of the Function field.

Arguments字段：可选字段，是对Function的补充，是指令在执行时对应的参数。这些参数可以包含流、服务或任何其他相关的信息。Arguments field: an optional field, which is a supplement to Function and is the parameter corresponding to the instruction when it is executed. These parameters can contain streams, services, or any other relevant information.

SRv6的每个Segment是128bit，可以灵活分为多段，每段功能和长度可以自定义，由此具备灵活编程能力，即业务可编辑。Each segment of SRv6 is 128bit, which can be flexibly divided into multiple segments, and the function and length of each segment can be customized, so it has flexible programming capabilities, that is, the business can be edited.

算力路由控制器101，用于基于算力服务信息和算力路由信息，确定各算力服务网关的算力服务SRv6 SID、多条算力路由路径及各算力接入网关的第一转发动作策略和各算力服务网关的第二转发动作策略；向每一算力接入网关下发每一算力接入网关的第一转发动作策略；向每一算力服务网关下发每一算力服务网关的算力服务SRv6 SID和第二转发动作策略。The computing power routing controller 101 is configured to determine, based on the computing power service information and the computing power routing information, the computing power service SRv6 SID of each computing power service gateway, multiple computing power routing paths, and the first forwarding of each computing power access gateway The action policy and the second forwarding action policy of each computing power service gateway; the first forwarding action policy of each computing power access gateway is delivered to each computing power access gateway; The computing power service SRv6 SID of the computing power service gateway and the second forwarding action policy.

其中，算力服务信息，包括各算力服务实例节点的负载状态信息、服务能力信息、提供的算力服务的算力服务ID和任播IP地址及连接的算力服务网关的IP地址；算力路由信息，包括各算力接入网关、各中继路由节点和各算力服务网关的拓扑信息、链路带宽信息、链路时延信息及链路丢包率信息；算力路由路径，指从一个算力接入网关，经至少一个中继路由节点到一个算力服务网关的路径；目标算力服务的任播IPv6地址，是基于网络服务提供商分配给目标算力服务的算力服务提供商的第一信息、目标算力服务的ID和补充参数确定的；算力服务网关的算力服务SRv6 SID，是基于网络服务提供商分配给算力服务网关的第二信息、与算力服务网关通信连接的各算力服务实例节点提供的算力服务的ID和补充参数确定的。Among them, the computing power service information, including the load status information of each computing power service instance node, service capability information, the computing power service ID and anycast IP address of the computing power service provided, and the IP address of the connected computing power service gateway; Power routing information, including topology information, link bandwidth information, link delay information, and link packet loss rate information of each computing power access gateway, each relay routing node, and each computing power service gateway; computing power routing paths, Refers to the path from a computing power access gateway to a computing power service gateway through at least one relay routing node; the anycast IPv6 address of the target computing power service is based on the computing power allocated to the target computing power service by the network service provider The first information of the service provider, the ID of the target computing power service and the supplementary parameters are determined; the computing power service SRv6 SID of the computing power service gateway is based on the second information allocated to the computing power service gateway by the network service provider, and the computing power service gateway. It is determined by the ID and supplementary parameters of the computing power service provided by each computing power service instance node that is communicatively connected to the power service gateway.

具体地，算力路由控制器101还可以用于获取算力服务信息和算力路由信息。Specifically, the computing power routing controller 101 can also be used to obtain computing power service information and computing power routing information.

可选地，算力路由控制器101，可以具体用于基于与算力服务编排器之间的通信接口，从算力服务编排器，实时获取所有的算力服务信息。Optionally, the computing power routing controller 101 may be specifically configured to acquire all computing power service information from the computing power service orchestrator in real time based on the communication interface with the computing power service orchestrator.

可选地，算力路由控制器101，还可以从其他设备实时获取所有的算力服务信息。Optionally, the computing power routing controller 101 can also acquire all computing power service information in real time from other devices.

算力服务信息，可以包括各算力服务实例节点的负载状态信息、各算力服务实例节点的服务能力信息、各算力服务实例节点的提供的算力服务的算力服务ID、各算力服务实例节点的提供的算力服务的任播IP地址，以及各算力服务实例节点连接的算力服务网关的IP地址。Computing service information, which may include load status information of each computing service instance node, service capability information of each computing service instance node, computing service ID of the computing service provided by each computing service instance node, and computing service ID of each computing service instance node. The anycast IP address of the computing service provided by the service instance node, and the IP address of the computing service gateway connected to each computing service instance node.

需要说明的是，算力路由控制器101获取算力服务信息之前，当算力服务编排器进行算力服务实例节点的生成时，为所有提供相同算力服务的各算力服务实例节点分配相同的该算力服务的任播IPv6地址，并对各算力服务实例节点进行注册。每一算力服务实例节点，均基于该算力服务实例节点所提供的算力服务的任播IPv6地址，与用户设备进行通信，从而实现用户设备对该算力服务和该算力服务实例节点的调用。It should be noted that, before the computing power routing controller 101 obtains the computing power service information, when the computing power service orchestrator generates the computing power service instance nodes, all the computing power service instance nodes that provide the same computing power service are allocated the same value. the anycast IPv6 address of the computing service, and register each computing service instance node. Each computing power service instance node communicates with the user equipment based on the anycast IPv6 address of the computing power service provided by the computing power service instance node, so as to realize the computing power service of the user equipment and the computing power service instance node. call.

可选地，算力服务实例节点属于逻辑实体或物理实体，可以由包括但不限于计算机、服务器、智能终端、虚拟机或容器等各类虚拟或物理设备承载其逻辑功能。Optionally, the computing power service instance node belongs to a logical entity or a physical entity, and its logical functions can be carried by various virtual or physical devices including but not limited to computers, servers, intelligent terminals, virtual machines or containers.

对于任一算力服务，该算力服务的任播IPv6地址，可以是基于网络服务提供商分配给该算力服务的算力服务提供商的第一信息、该算力服务的ID和补充参数确定的。算力服务的任播IPv6地址，可以用于标识提供该算力服务的算力服务实例节点。For any computing power service, the anycast IPv6 address of the computing power service can be based on the first information of the computing power service provider assigned by the network service provider to the computing power service, the ID of the computing power service, and the supplementary parameters definite. The anycast IPv6 address of the computing service can be used to identify the computing service instance node that provides the computing service.

算力服务的任播IPv6地址包括前缀和后缀。其中，前缀由网络服务提供商(Internet Service Provider，ISP)分配给算力服务提供商(Computing ServiceProvider，CSP)。ISP分配给CSP前缀，可以称为第一信息。该第一信息，可以用于标识该CSP。ISP分配给不同的CSP的前缀也不同，即前缀可以是CSP在全局的唯一标识。后缀可以由CSP分配。CSP分配的后缀可以包括算力服务的ID及补充参数两部分。The anycast IPv6 address of the computing service includes a prefix and a suffix. The prefix is allocated by an Internet Service Provider (ISP) to a Computing Service Provider (CSP). The prefix assigned by the ISP to the CSP may be referred to as the first information. The first information may be used to identify the CSP. The prefixes assigned by the ISP to different CSPs are also different, that is, the prefixes can be the globally unique identifiers of the CSPs. Suffixes can be assigned by the CSP. The suffix allocated by the CSP may include the ID of the computing power service and the supplementary parameters.

算力服务的ID，用于标识算力服务，是某一算力服务在全局的唯一标识。所有提供同一算力服务的不同算力服务实例节点都使用同一个通过前述算力服务任播IP地址分配方法产生的算力服务的任播IPv6地址与用户设备进行通信；而所有提供不同算力服务的算力服务实例节点都使用不同的算力服务的任播IPv6地址与用户进行通信。The ID of the computing service is used to identify the computing service, and is the unique identifier of a computing service globally. All different computing power service instance nodes that provide the same computing power service use the same anycast IPv6 address of the computing power service generated by the aforementioned computing power service anycast IP address allocation method to communicate with user equipment; and all nodes that provide different computing power The computing service instance nodes of the service all use the anycast IPv6 addresses of different computing services to communicate with users.

可选地，算力路由控制器101，可以具体用于通过与各算力接入网关102、各中继路由节点103(或SDN路由控制器)和各算力服务网关104之间的通信接口，从各算力接入网关102、各中继路由节点103(或SDN路由控制器)和各算力服务网关104，实时获取全网的算力路由信息。Optionally, the computing power routing controller 101 can be specifically used to pass the communication interface with each computingpower access gateway 102, each relay routing node 103 (or SDN routing controller) and each computingpower service gateway 104. , from each computingpower access gateway 102 , each relay routing node 103 (or SDN routing controller) and each computingpower service gateway 104 , obtain the computing power routing information of the entire network in real time.

可选地，算力路由控制器101，还可以从其他设备实时获取所有的算力路由信息。Optionally, the computing power routing controller 101 may also acquire all computing power routing information in real time from other devices.

算力路由信息，可以包括各算力接入网关102、各中继路由节点103和各算力服务网关104作为节点构成的网络的拓扑信息，每一算力接入网关102和与该算力接入网关102通信连接的每一中继路由节点103之间的通信链路的链路带宽信息、链路时延信息及链路丢包率信息，每一中继路由节点103和与该中继路由节点103通信连接的每一其他中继路由节点103之间的通信链路的链路带宽信息、链路时延信息及链路丢包率信息，以及每一中继路由节点103和与该中继路由节点103通信连接的每一算力服务网关104之间的通信链路的链路带宽信息、链路时延信息及链路丢包率信息。The computing power routing information may include the topology information of the network formed by each computingpower access gateway 102, eachrelay routing node 103 and each computingpower service gateway 104 as nodes, each computingpower access gateway 102 and the computing power The link bandwidth information, link delay information and link packet loss rate information of the communication link between eachrelay routing node 103 that theaccess gateway 102 communicates with, eachrelay routing node 103 and the middle Link bandwidth information, link delay information, and link packet loss rate information of the communication link between each otherrelay routing node 103 that is communicatively connected to therouting node 103, and eachrelay routing node 103 and the Link bandwidth information, link delay information, and link packet loss rate information of the communication link between each computingpower service gateway 104 that therelay routing node 103 communicates with.

算力路由控制器101，可以具体用于根据实时获取到的算力服务信息和算力路由信息，为各算力接入网关102规划多条到能够满足算力服务资源需求的算力服务实例节点所连接的算力服务网关104的算力路由路径。The computing power routing controller 101 can be specifically used to plan multiple computing power service instances for each computingpower access gateway 102 to meet the computing power service resource requirements according to the computing power service information and computing power routing information obtained in real time The computing power routing path of the computingpower service gateway 104 to which the node is connected.

可以理解的是，每一算力服务实例节点与至少一个算力服务网关104通信连接，因而每一条算力路由路径，均以一个算力接入网关102为起点，经过一个或多个中继路由节点103，并以一个算力服务网关104为终点。It can be understood that each computing power service instance node is in communication connection with at least one computingpower service gateway 104, so each computing power routing path starts from a computingpower access gateway 102 and passes through one or more relays. Therouting node 103 ends with a computingpower service gateway 104 .

当算力服务实例节点通过算力服务编排器完成服务生成及注册流程后，算力路由控制器101可以通过与算力服务编排器之间的通信接口，获取该算力服务实例节点所提供的算力服务的算力服务ID、该算力服务任播IP地址及连接的算力服务网关104的IP地址(可以理解的是，上述信息为算力服务信息中的一部分)；基于上述信息，确定该算力服务网关104的算力服务SRv6 SID，并向该算力服务网关104下发该算力服务网关104的算力服务SRv6 SID。After the computing power service instance node completes the service generation and registration process through the computing power service orchestrator, the computing power routing controller 101 can obtain the information provided by the computing power service instance node through the communication interface with the computing power service orchestrator. The computing power service ID of the computing power service, the anycast IP address of the computing power service, and the IP address of the connected computing power service gateway 104 (it can be understood that the above information is a part of the computing power service information); based on the above information, The computing power service SRv6 SID of the computingpower service gateway 104 is determined, and the computing power service SRv6 SID of the computingpower service gateway 104 is delivered to the computingpower service gateway 104 .

需要说明的是，算力服务网关104可以与至少一个算力服务实例节点通信连接，该算力服务网关104连接的算力服务实例节点提供的算力服务可以相同也可以不同，因而该算力服务网关104的算力服务SRv6 SID可以有多个，其数量等于与该算力服务网关104连接的算力服务实例节点提供的算力服务的数量。It should be noted that the computingpower service gateway 104 may be connected to at least one computing power service instance node, and the computing power service provided by the computing power service instance node connected to the computingpower service gateway 104 may be the same or different. Therefore, the computing power There may be multiple computing power service SRv6 SIDs of theservice gateway 104 , the number of which is equal to the number of computing power services provided by the computing power service instance nodes connected to the computingpower service gateway 104 .

对于任一算力服务网关104的任一算力服务SRv6 SID，可以是基于网络服务提供商分配给算力服务网关的第二信息、与算力服务网关通信连接的算力服务实例节点提供的某个算力服务的ID和补充参数确定的。For any computing power service SRv6 SID of any computingpower service gateway 104, it may be provided by the computing power service instance node communicatively connected to the computing power service gateway based on the second information allocated to the computing power service gateway by the network service provider It is determined by the ID and supplementary parameters of a computing service.

任一算力服务网关104的任一算力服务SRv6 SID中的Locator字段，可以是ISP分配给该算力服务网关104的第二信息。该第二信息，可以用于标识该算力服务网关104。ISP分配给不同的算力服务网关104的第二信息也不同，即第二信息可以是算力服务网关104在全局的唯一标识。该算力服务SRv6 SID中的Function及Argument字段，可以由CSP分配给该算力服务网关104。CSP分配的Function字段对应与该算力服务网关104通信连接的算力服务实例节点提供的某个算力服务的ID；CSP分配的Argument字段对应上述某个算力服务的补充参数。The Locator field in any computing power service SRv6 SID of any computingpower service gateway 104 may be the second information allocated to the computingpower service gateway 104 by the ISP. The second information may be used to identify the computingpower service gateway 104 . The second information allocated by the ISP to different computingpower service gateways 104 is also different, that is, the second information may be the globally unique identifier of the computingpower service gateway 104 . The Function and Argument fields in the computing power service SRv6 SID can be allocated to the computingpower service gateway 104 by the CSP. The Function field allocated by the CSP corresponds to the ID of a certain computing power service provided by the computing power service instance node communicatively connected to the computingpower service gateway 104; the Argument field allocated by the CSP corresponds to the supplementary parameter of the above-mentioned certain computing power service.

算力服务的任播IPv6地址，可以用于标识提供该算力服务的算力服务实例节点。算力服务网关104的算力服务SRv6 SID，可以用于该标识算力服务网关后端(即连接的算力服务实例节点)承载的算力服务。基于算力服务的任播IPv6地址和算力服务网关104的算力服务SRv6 SID，可以便于通过SRv6网络控制平面实现对算力服务实例节点调用关系的编排控制。The anycast IPv6 address of the computing service can be used to identify the computing service instance node that provides the computing service. The computing power service SRv6 SID of the computingpower service gateway 104 can be used to identify the computing power service carried by the back end of the computing power service gateway (ie, the connected computing power service instance node). Based on the anycast IPv6 address of the computing power service and the computing power service SRv6 SID of the computingpower service gateway 104, it is convenient to realize the orchestration control of the calling relationship of the computing power service instance nodes through the SRv6 network control plane.

每一算力服务网关104，基于通过前述算力服务SRv6 SID分配方法产生的算力服务SRv6 SID，与算力路由控制器101、算力接入网关102、中继路由节点103及其他算力服务网关104进行交互。Each computingpower service gateway 104, based on the computing power service SRv6 SID generated by the aforementioned computing power service SRv6 SID allocation method, communicates with the computing power routing controller 101, the computingpower access gateway 102, therelay routing node 103 and other computing power. Theservice gateway 104 interacts.

算力路由控制器101可以基于规划确定的每一条算力路由路径，确定该条算力路由路径经过的算力接入网关102的第一转发动作策略和算力服务网关104的第二转发动作策略。算力接入网关102的第一转发动作策略，用于使该算力接入网关102执行对应的转发动作，实现通过该条算力路由路径的流量转发。算力服务网关104的第二转发动作策略，用于使该算力服务网关104执行对应的转发动作，将流量转发至与该算力服务网关104连接的对应的算力服务实例节点。The computing power routing controller 101 may, based on each computing power routing path determined in the plan, determine the first forwarding action policy of the computingpower access gateway 102 and the second forwarding action of the computingpower service gateway 104 through which the computing power routing path passes. Strategy. The first forwarding action policy of the computingpower access gateway 102 is used to enable the computingpower access gateway 102 to perform a corresponding forwarding action to implement traffic forwarding through the computing power routing path. The second forwarding action policy of the computingpower service gateway 104 is used to make the computingpower service gateway 104 perform a corresponding forwarding action to forward traffic to the corresponding computing power service instance node connected to the computingpower service gateway 104 .

通过上述步骤，算力路由控制器101可以确定各算力接入网关102的第一转发动作策略和各算力服务网关104的第二转发动作策略。确定各算力接入网关102的第一转发动作策略和各算力服务网关104的第二转发动作策略之后，算力路由控制器101向每一算力接入网关102下发该算力接入网关102的第一转发动作策略，并向每一算力服务网关104下发该算力服务网关104的第二转发动作策略。Through the above steps, the computing power routing controller 101 can determine the first forwarding action policy of each computingpower access gateway 102 and the second forwarding action policy of each computingpower service gateway 104 . After determining the first forwarding action policy of each computingpower access gateway 102 and the second forwarding action policy of each computingpower service gateway 104, the computing power routing controller 101 sends the computingpower access gateway 102 to each computingpower access gateway 102. the first forwarding action policy of theingress gateway 102 , and deliver the second forwarding action policy of the computingpower service gateway 104 to each computingpower service gateway 104 .

可选地，中继路由节点103的第三转发策略，用于使该中继路由节点103执行对应的转发动作，实现通过该中继路由节点103的算力路由路径的流量转发。Optionally, the third forwarding policy of therelay routing node 103 is used to enable therelay routing node 103 to perform a corresponding forwarding action to implement traffic forwarding through the computing power routing path of therelay routing node 103 .

可选地，中继路由节点103的第三转发策略，可以由算力路由控制器101确定并下发，也可以预先配置在该中继路由节点103的本地存储中，还可以由SDN路由控制器确定并下发。Optionally, the third forwarding policy of therelay routing node 103 may be determined and issued by the computing power routing controller 101, or may be pre-configured in the local storage of therelay routing node 103, and may also be controlled by SDN routing. The device is determined and issued.

算力接入网关102，用于基于用户设备105发送的算力业务流量携带的目标算力服务的任播IPv6地址和算力接入网关102的第一转发动作策略，将算力业务流量封装对应的SR扩展头之后，将算力业务流量转发至相应的中继路由节点103。The computingpower access gateway 102 is configured to encapsulate the computing power service traffic based on the anycast IPv6 address of the target computing power service carried by the computing power service traffic sent by theuser equipment 105 and the first forwarding action policy of the computingpower access gateway 102 After the corresponding SR header is extended, the computing power service traffic is forwarded to the correspondingrelay routing node 103 .

具体地，用户设备105用于发起算力服务的调用，向该算力路由系统发送算力业务流量，以通过该算力路由系统进行路由，将该算力业务流量转发至目标算力服务实例节点。Specifically, theuser equipment 105 is used to initiate the call of the computing power service, send the computing power service traffic to the computing power routing system for routing through the computing power routing system, and forward the computing power service flow to the target computing power service instance node.

本发明实施例中，用户设备在调用算力服务时，不是通过域名来标识算力服务的地址，而通过算力服务的任播IP地址来直接调用算力服务。相应地，算力业务流量携带的目的地址为目标算力服务的任播IPv6地址。目标算力服务，指待调用的算力业务流量。目标算力服务的数量，可以为一个或多个。In the embodiment of the present invention, when invoking the computing power service, the user equipment does not use the domain name to identify the address of the computing power service, but directly calls the computing power service through the anycast IP address of the computing power service. Correspondingly, the destination address carried by the computing power service traffic is the anycast IPv6 address of the target computing power service. The target computing power service refers to the computing power business traffic to be invoked. The number of target computing services, which can be one or more.

算力接入网关102接收算力业务流量之后，可以将算力业务流量携带的目标算力服务的任播IPv6地址与自身的第一转发动作策略中的算力服务SRv6 SID表项进行匹配；基于与目标算力服务的任播IPv6地址匹配的算力服务SRv6 SID，确定实际的转发路径，并为该算力业务流量封装对应的SRH(指SR扩展头)，将实际的转发路径的各跳节点的SID封装在对应的SRH中，然后将封装对应的SRH的算力业务流量转发至相应的中继路由节点103。After the computingpower access gateway 102 receives the computing power service traffic, it can match the anycast IPv6 address of the target computing power service carried by the computing power service traffic with the computing power service SRv6 SID entry in its own first forwarding action policy; Based on the computing service SRv6 SID matching the anycast IPv6 address of the target computing service, the actual forwarding path is determined, and the corresponding SRH (referring to the SR extension header) is encapsulated for the computing service traffic. The SID of the hop node is encapsulated in the corresponding SRH, and then the computing power service traffic encapsulated in the corresponding SRH is forwarded to the correspondingrelay routing node 103 .

相应的中继路由节点103，是实际的转发路径中该算力接入网关102的下一跳节点。The correspondingrelay routing node 103 is the next hop node of the computingpower access gateway 102 in the actual forwarding path.

实际的转发路径，可以为起点为该算力接入网关102且终点为算力服务SRv6 SID是与目标算力服务的任播IPv6地址匹配的算力服务SRv6 SID的算力服务网关104的算力路由路径。该算力接入网关102，可以从起点为该算力接入网关102且终点为算力服务SRv6SID是与目标算力服务的任播IPv6地址匹配的算力服务SRv6 SID的算力服务网关104的所有算力路由路径中选择一条或多条，作为实际的转发路径。The actual forwarding path can be the computingpower service gateway 104 whose origin is the computingpower access gateway 102 and the end point is the computing power service SRv6 SID which is the computing power service SRv6 SID that matches the anycast IPv6 address of the target computing power service. force routing path. The computingpower access gateway 102 can start from the computingpower access gateway 102 and end at the computingpower service gateway 104 whose SRv6 SID is the computing power service SRv6 SID matching the anycast IPv6 address of the target computing power service Select one or more of all computing power routing paths as the actual forwarding path.

可以理解的是，上述实际的转发路径的终点可以有多个，因而实际的转发路径可以有多条。上述实际的转发路径的终点也可以只有一个，在述实际的转发路径的终点只有一个的情况下，实际的转发路径也可以有多条。It can be understood that, there may be multiple end points of the above-mentioned actual forwarding path, and thus there may be multiple actual forwarding paths. There may be only one end point of the above-mentioned actual forwarding path. In the case where there is only one end point of the actual forwarding path, there may be multiple actual forwarding paths.

中继路由节点103，用于基于中继路由节点103的第三转发动作策略和对应的SR扩展头，将接收的封装对应的SR扩展头的算力业务流量进行转发。Therelay routing node 103 is configured to forward the received computing power service traffic encapsulated with the corresponding SR extension header based on the third forwarding action policy of therelay routing node 103 and the corresponding SR extension header.

具体地，接收到封装对应的SR扩展头的算力业务流量的任一中继路由节点103，可以基于自身的第三转发动作策略，根据对应的SR扩展头中下一跳节点的SRv6 SID，将封装对应的SR扩展头的算力业务流量转发至下一跳节点。下一跳节点，可以为另一中继路由节点103或某一算力服务网关104。Specifically, anyrelay routing node 103 that receives the computing power service traffic encapsulating the corresponding SR extension header may, based on its own third forwarding action policy, and according to the SRv6 SID of the next hop node in the corresponding SR extension header, Forward the computing power service traffic encapsulated with the corresponding SR extension header to the next hop node. The next hop node may be anotherrelay routing node 103 or a certain computingpower service gateway 104 .

算力服务网关104，用于基于算力服务网关104的第二转发动作策略、对应的SR扩展头和算力服务网关104的算力服务SRv6 SID，将算力业务流量转发至目标算力服务实例节点。The computingpower service gateway 104 is used to forward the computing power service traffic to the target computing power service based on the second forwarding action policy of the computingpower service gateway 104, the corresponding SR extension header and the computing power service SRv6 SID of the computingpower service gateway 104 instance node.

其中。目标算力服务实例节点，包括至少一个与算力服务网关通信连接的算力服务实例节点。in. The target computing power service instance node includes at least one computing power service instance node that is communicatively connected to the computing power service gateway.

具体地，接收到封装对应的SR扩展头的算力业务流量的算力服务网关104，可以将SR扩展头中的目的地址，与自身的算力服务SRv6 SID进行匹配；基于自身的第二转发动作策略，基于与目的地址匹配的一个算力服务SRv6 SID，剥离对应的SR扩展头，将剥离对应的SR扩展头后的算力业务流量转发至目标算力服务实例节点。Specifically, the computingpower service gateway 104 that receives the computing power service traffic encapsulating the corresponding SR extension header can match the destination address in the SR extension header with its own computing power service SRv6 SID; based on its own second forwarding The action policy, based on a computing service SRv6 SID matching the destination address, strips the corresponding SR extension header, and forwards the computing service traffic after stripping the corresponding SR extension header to the target computing service instance node.

目标算力服务实例节点，是提供的算力服务的SRv6 SID目的地址匹配的、与该算力服务网关104连接的一个或多个算力服务实例节点。The target computing power service instance node is one or more computing power service instance nodes connected to the computingpower service gateway 104 that match the SRv6 SID destination address of the provided computing power service.

本发明实施例基于算力服务的任播IP地址和算力服务网关的算力服务SRv6 SID进行算力业务流量的转发，对算力网络的基础设施的改动更小，能够在对算力网络的基础设施较小改动的前提下，实现算力网络资源更高效的调用，能通过算力路由控制器实现了对算力服务的高效调用，即“以网取算”。The embodiment of the present invention forwards the computing power service traffic based on the anycast IP address of the computing power service and the computing power service SRv6 SID of the computing power service gateway, and the changes to the infrastructure of the computing power network are smaller. On the premise of minor changes to the infrastructure, more efficient invocation of computing power network resources can be achieved, and efficient invocation of computing power services can be achieved through the computing power routing controller, that is, "taking calculations through the network".

基于上述任一实施例的内容，算力路由控制器101，还用于基于算力服务信息和算力路由信息，确定每一条算力路由路径的优先级和/或权重；算力接入网关的第一转发动作策略，包括经过算力接入网关的算力路由路径的优先级和/或权重。Based on the content of any of the above embodiments, the computing power routing controller 101 is further configured to determine the priority and/or weight of each computing power routing path based on the computing power service information and the computing power routing information; the computing power access gateway The first forwarding action strategy includes the priority and/or weight of the computing power routing path through the computing power access gateway.

具体地，算力路由控制器101在基于算力服务信息和算力路由信息确定算力路由路径时，还可以确定每一条算力路由路径的优先级和/或权重。Specifically, when determining the computing power routing path based on the computing power service information and the computing power routing information, the computing power routing controller 101 may also determine the priority and/or weight of each computing power routing path.

算力路由路径的优先级，用于算力接入网关102在选择实际的转发路径时，按照算力路由路径的优先级由高到低的顺序来确定实际的转发路径。The priority of the computing power routing path is used for the computingpower access gateway 102 to determine the actual forwarding path in descending order of the priority of the computing power routing path when selecting the actual forwarding path.

可选地，算力路由路径的优先级越高，经过的节点的链路带宽信息、链路时延信息及链路丢包率信息的综合性能更佳。Optionally, the higher the priority of the computing power routing path, the better the comprehensive performance of the link bandwidth information, the link delay information and the link packet loss rate information of the passing nodes.

可选地，在链路带宽和链路时延接近的情况下，链路丢包率越小的算力路由路径的优先级越高，链路丢包率越大的算力路由路径的优先级越低；在链路带宽和链路丢包率接近的情况下，链路时延越小的算力路由路径的优先级越高，链路时延越大的算力路由路径的优先级越低；在链路时延及链路丢包率接近的情况下，链路带宽越大的算力路由路径的优先级越高，链路带宽越小的算力路由路径的优先级越低。Optionally, when the link bandwidth and link delay are close, the lower the link packet loss rate is, the higher the priority of the computing power routing path, and the higher the link packet loss rate is, the higher the priority of the computing power routing path. The lower the level; when the link bandwidth and link packet loss rate are close, the smaller the link delay is, the higher the priority of the computing power routing path, and the higher the priority of the computing power routing path with the larger link delay The lower the value; when the link delay and link packet loss rate are close, the higher the priority of the computing power routing path with the larger link bandwidth, the lower the priority of the computing power routing path with the smaller link bandwidth .

算力路由路径的权重，用于确定各实际的转发路径承担的算力业务流量的比例。在多条实际的转发路径中，权重越大的实际的转发路径承担的比例越大，权重越小的实际的转发路径承担的比例越小，权重相同的实际的转发路径承担的比例可以相同。The weight of the computing power routing path is used to determine the proportion of computing power service traffic borne by each actual forwarding path. Among the multiple actual forwarding paths, the actual forwarding paths with larger weights have a larger proportion, and the smaller the weights, the smaller the actual forwarding paths. The actual forwarding paths with the same weight can have the same proportion.

本发明实施例通过算力路由控制器基于算力服务信息和算力路由信息，确定每一条算力路由路径的优先级和/或权重，能实现更高效的对算力服务的调用。In the embodiment of the present invention, the computing power routing controller determines the priority and/or weight of each computing power routing path based on the computing power service information and the computing power routing information, so that more efficient invocation of the computing power service can be realized.

基于上述任一实施例的内容，算力路由控制器101，还用于基于新的算力服务信息和新的算力路由信息，确定上一次下发的各第一转发动作策略和各第二转发动作策略是否满足需求；在不满足需求的情况下，基于新的算力服务信息和新的算力路由信息，重新确定多条算力路由路径及各算力接入网关的第一转发动作策略和各算力服务网关的转发动作策略；向每一算力接入网关下发重新确定的每一算力接入网关的转发动作策略；向每一算力服务网关下发重新确定的每一算力服务网关的第二转发动作策略。Based on the content of any of the above embodiments, the computing power routing controller 101 is further configured to determine, based on the new computing power service information and the new computing power routing information, each first forwarding action policy and each second forwarding action strategy issued last time Whether the forwarding action strategy meets the requirements; if the requirements are not met, based on the new computing power service information and new computing power routing information, re-determine multiple computing power routing paths and the first forwarding action of each computing power access gateway strategy and forwarding action strategy of each computing power service gateway; deliver the re-determined forwarding action policy of each computing power access gateway to each computing power access gateway; A second forwarding action policy of the computing power service gateway.

具体地，算力路由控制器101可以实时获取算力服务信息和算力路由信息，因而可以根据所获取的新的算力服务信息和新的算力路由信息，判断上一次确定并下发的各第一转发动作策略和各第二转发动作策略(在第三转发动作策略由算力路由控制器101确定的情况下，还可以包括第三转发动作策略)是否满足算力服务资源的需求。Specifically, the computing power routing controller 101 can obtain the computing power service information and the computing power routing information in real time, and thus can judge the last determined and issued data according to the acquired new computing power service information and new computing power routing information. Whether each first forwarding action strategy and each second forwarding action strategy (if the third forwarding action strategy is determined by the computing power routing controller 101, may also include a third forwarding action strategy) meet the requirements of computing power service resources.

如果不满足，则算力路由控制器101可以根据新的算力服务信息和新的算力路由信息，重新规划算力路由路径，重新确定各第一转发动作策略和各第二转发动作策略(在第三转发动作策略由算力路由控制器101确定的情况下，还可以包括第三转发动作策略)并下发。If not, the computing power routing controller 101 can re-plan the computing power routing path according to the new computing power service information and the new computing power routing information, and re-determine each first forwarding action strategy and each second forwarding action strategy ( In the case where the third forwarding action policy is determined by the computing power routing controller 101, the third forwarding action policy may also be included and delivered.

可选地，在算力路由控制器101还基于算力服务信息和算力路由信息，确定每一条算力路由路径的优先级和/或权重的情况下，重新规划算力路由路径时，相应地重新确定每一条算力路由路径的优先级和/或权重。Optionally, when the computing power routing controller 101 also determines the priority and/or weight of each computing power routing path based on the computing power service information and the computing power routing information, when re-planning the computing power routing path, the corresponding to re-determine the priority and/or weight of each computing power routing path.

如果满足，则可以保持上一次确定的算力路由路径，各第一转发动作策略和各第二转发动作策略(在第三转发动作策略由算力路由控制器101确定的情况下，还可以包括第三转发动作策略)。If satisfied, the computing power routing path determined last time can be maintained, each first forwarding action policy and each second forwarding action policy (in the case that the third forwarding action policy is determined by the computing power routing controller 101, it can also include The third forwarding action policy).

本发明实施例通过在上一次下发的各第一转发动作策略和各第二转发动作策略不满足需求的情况下，基于新的算力服务信息和新的算力路由信息，重新确定多条算力路由路径及各算力接入网关的第一转发动作策略和各算力服务网关的转发动作策略并下发，能实现更高效的对算力服务的调用。In this embodiment of the present invention, when each first forwarding action policy and each second forwarding action policy issued last time do not meet the requirements, based on new computing power service information and new computing power routing information, re-determine multiple The computing power routing path and the first forwarding action policy of each computing power access gateway and the forwarding action policy of each computing power service gateway are issued concurrently, which can achieve more efficient invocation of computing power services.

基于上述任一实施例的内容，算力路由控制器101，还用于基于算力服务信息和算力路由信息，确定各中继路由节点的第三转发动作策略；向每一中继路由节点下发每一中继路由节点的第三转发动作策略。Based on the content of any of the above embodiments, the computing power routing controller 101 is further configured to determine the third forwarding action strategy of each relay routing node based on the computing power service information and the computing power routing information; The third forwarding action policy of each relay routing node is delivered.

具体地，算力路由控制器101还可以在确定各条算力路由路径之后，可以确定每一条算力路由路径经过的各中继路由节点103的第三转发动作策略。Specifically, after determining each computing power routing path, the computing power routing controller 101 may also determine the third forwarding action policy of eachrelay routing node 103 that each computing power routing path passes through.

通过上述步骤，算力路由控制器101可以确定各中继路由节点103的第三转发动作策略，然后可以向每一中继路由节点103下发该中继路由节点103的第三转发动作策略。Through the above steps, the computing power routing controller 101 can determine the third forwarding action policy of eachrelay routing node 103 , and then can deliver the third forwarding action policy of therelay routing node 103 to eachrelay routing node 103 .

本发明实施例通过算力路由控制器基于算力服务信息和算力路由信息，确定各中继路由节点的第三转发动作策略并下发至该中继路由节点，能实现基于算力服务的任播IP地址和算力服务网关的算力服务SRv6 SID进行算力业务流量的转发，从而能够在对算力网络的基础设施较小改动的前提下，实现算力网络资源更高效的调用。In this embodiment of the present invention, the computing power routing controller determines the third forwarding action strategy of each relay routing node based on the computing power service information and the computing power routing information and sends it to the relay routing node, so that the computing power service-based The anycast IP address and the computing power service SRv6 SID of the computing power service gateway forward the computing power service traffic, so that the computing power network resources can be called more efficiently under the premise of minor changes to the computing power network infrastructure.

基于上述任一实施例的内容，算力路由控制器101，还用于从算力服务编排器，获取算力服务信息。Based on the content of any of the above embodiments, the computing power routing controller 101 is further configured to obtain computing power service information from the computing power service orchestrator.

具体地，算力路由控制器101，可以具体用于基于与算力服务编排器之间的通信接口，从算力服务编排器，实时获取所有的算力服务信息。Specifically, the computing power routing controller 101 can be specifically configured to obtain all computing power service information in real time from the computing power service orchestrator based on the communication interface with the computing power service orchestrator.

基于上述任一实施例的内容，各算力接入网关102、各中继路由节点103和各算力服务网关104，还用于维护链路状态测量会话，以获取链路时延信息及链路丢包率信息。Based on the content of any of the above embodiments, each computingpower access gateway 102, eachrelay routing node 103 and each computingpower service gateway 104 are also used to maintain link state measurement sessions to obtain link delay information and link status. Packet loss rate information.

具体地，每一算力接入网关102，可以维护与其他节点(可以包括与该算力接入网关102连接的各中继路由节点103)间的链路状态测量会话，从而可以获取该链路实时的链路时延信息及链路丢包率信息。Specifically, each computingpower access gateway 102 can maintain a link state measurement session with other nodes (which may include eachrelay routing node 103 connected to the computing power access gateway 102), so as to obtain the link status measurement session. Real-time link delay information and link packet loss rate information.

每一中继路由节点103，可以维护与其他节点(可以包括与该中继路由节点103连接的各其他中继路由节点103、算力接入网关102和算力服务网关104)间的链路状态测量会话，从而可以获取该链路实时的链路时延信息及链路丢包率信息。Eachrelay routing node 103 can maintain links with other nodes (which may include otherrelay routing nodes 103, computingpower access gateway 102 and computingpower service gateway 104 connected to the relay routing node 103). Status measurement session, so that the real-time link delay information and link packet loss rate information of the link can be obtained.

每一算力服务网关104，可以维护与其他节点(可以包括与该算力服务网关104连接的各中继路由节点103和各其他算力服务网关104)间的链路状态测量会话，从而可以获取该链路实时的链路时延信息及链路丢包率信息。Each computingpower service gateway 104 can maintain a link state measurement session with other nodes (which may include eachrelay routing node 103 connected to the computingpower service gateway 104 and each other computing power service gateway 104), so that it can Obtain the real-time link delay information and link packet loss rate information of the link.

算力路由控制器101，还用于从各算力接入网关、各中继路由节点和各算力服务网关，获取算力路由信息。The computing power routing controller 101 is further configured to obtain computing power routing information from each computing power access gateway, each relay routing node, and each computing power service gateway.

具体地，算力路由控制器101，可以具体用于通过与各算力接入网关102、各中继路由节点103(或SDN路由控制器)和各算力服务网关104之间的通信接口，从各算力接入网关102、各中继路由节点103(或SDN路由控制器)和各算力服务网关104，实时获取全网的算力路由信息。Specifically, the computing power routing controller 101 can be specifically used to communicate with each computingpower access gateway 102, each relay routing node 103 (or SDN routing controller), and each computingpower service gateway 104. From each computingpower access gateway 102, each relay routing node 103 (or SDN routing controller) and each computingpower service gateway 104, the computing power routing information of the entire network is obtained in real time.

基于上述任一实施例的内容，算力接入网关102，还用于基于获取到的链路时延信息及链路丢包率信息，更新算力接入网关102的第一转发动作策略中的算力路由路径的优先级和/或权重。Based on the content of any one of the above embodiments, the computingpower access gateway 102 is further configured to update the first forwarding action policy of the computingpower access gateway 102 based on the acquired link delay information and link packet loss rate information The priority and/or weight of the computing power routing path.

具体地，算力接入网关102可以在基于自身维护的链路状态测量会话，获取链路时延信息及链路丢包率信息之后，自主地根据获取到的链路时延信息及链路丢包率信息，更新从算力路由控制器101获取的自身的第一转发动作策略中的各算力路由路径的优先级和/或权重。Specifically, the computingpower access gateway 102 may, after acquiring the link delay information and link packet loss rate information based on the link state measurement session maintained by itself, autonomously measure the link based on the acquired link delay information and link packet loss rate information. The packet loss rate information is used to update the priority and/or weight of each computing power routing path in its own first forwarding action policy obtained from the computing power routing controller 101 .

本发明实施例通过算力接入网关基于获取到的链路时延信息及链路丢包率信息，更新自身的第一转发动作策略中的算力路由路径的优先级和/或权重，能实现更高效的对算力服务的调用。This embodiment of the present invention enables the computing power access gateway to update the priority and/or weight of the computing power routing path in its own first forwarding action policy based on the acquired link delay information and link packet loss rate information, thereby enabling Realize more efficient calls to computing services.

基于上述任一实施例的内容，链路状态测量会话基于的协议包括Internet控制报文协议或双向主动测量协议。Based on the contents of any of the above embodiments, the protocol based on the link state measurement session includes the Internet Control Message Protocol or the Bidirectional Active Measurement Protocol.

具体地，Internet控制报文协议(Internet Control Message Protocol，ICMP，互联网控制报文协议)是TCP/IP协议簇的一个子协议，用于在IP主机、路由器之间传递控制消息。控制消息是指网络通不通、主机是否可达、路由是否可用等网络本身的消息。这些控制消息虽然并不传输用户数据，但是对于用户数据的传递起着重要的作用。Specifically, the Internet Control Message Protocol (ICMP, Internet Control Message Protocol) is a sub-protocol of the TCP/IP protocol suite, and is used to transmit control messages between IP hosts and routers. Control messages refer to the messages of the network itself, such as whether the network is reachable, whether the host is reachable, and whether the route is available. Although these control messages do not transmit user data, they play an important role in the transmission of user data.

双向主动测量协议(Tow-Way Active Measurement Protocol，TWAMP)是一种用于IP链路的性能测量技术，可以在正反两个方向进行双向的性能统计。TWAMP使用UDP数据包作为测量探帧，统计网络双向延迟、抖动。同时，该协议具有安全性，能够保证控制和测量功能的分离。通过已经部署TWAMP协议的网络设备之间的合作，可以有效地获取设备间IP性能统计数据。Two-way Active Measurement Protocol (Tow-Way Active Measurement Protocol, TWAMP) is a performance measurement technology for IP links, which can perform bidirectional performance statistics in both forward and reverse directions. TWAMP uses UDP data packets as measurement detection frames to count network two-way delay and jitter. At the same time, the protocol is secure and can guarantee the separation of control and measurement functions. Inter-device IP performance statistics can be efficiently obtained through cooperation between network devices that have deployed the TWAMP protocol.

下面对本发明提供的基于SRv6的算力路由方法进行描述，下文描述的基于SRv6的算力路由方法与上文描述的基于SRv6的算力路由系统可相互对应参照。The SRv6-based computing power routing method provided by the present invention will be described below. The SRv6-based computing power routing method described below and the SRv6-based computing power routing system described above may refer to each other correspondingly.

图2是本发明提供的基于SRv6的算力路由方法的流程示意图之一。如图2所示，本发明实施例提供的基于SRv6的算力路由方法的执行主体可以为基于SRv6的算力路由系统，该方法包括：步骤201、步骤202、步骤203和步骤204。FIG. 2 is one of the schematic flowcharts of the SRv6-based computing power routing method provided by the present invention. As shown in FIG. 2 , the execution body of the SRv6-based computing power routing method provided by the embodiment of the present invention may be an SRv6-based computing power routing system, and the method includes:step 201 ,step 202 , step 203 , and step 204 .

具体地，本发明实施例提供的基于SRv6的算力路由方法，是基于前述基于SRv6的算力路由系统实现的。Specifically, the SRv6-based computing power routing method provided by the embodiment of the present invention is implemented based on the aforementioned SRv6-based computing power routing system.

步骤201、在调用单个目标算力服务的情况下，算力路由控制器基于算力服务信息和算力路由信息，确定各算力服务网关的算力服务SRv6 SID、多条算力路由路径及各算力接入网关的第一转发动作策略和各算力服务网关的第二转发动作策略；向每一算力接入网关下发每一算力接入网关的第一转发动作策略；向每一算力服务网关下发每一算力服务网关的算力服务SRv6 SID和第二转发动作策略。Step 201: In the case of invoking a single target computing power service, the computing power routing controller determines the computing power service SRv6 SID, multiple computing power routing paths and the first forwarding action policy of each computing power access gateway and the second forwarding action policy of each computing power service gateway; delivering the first forwarding action policy of each computing power access gateway to each computing power access gateway; Each computing power service gateway delivers the computing power service SRv6 SID and the second forwarding action policy of each computing power service gateway.

具体地，在面向单算力服务调用(指调用单个目标算力服务)的情况下，算力路由控制器可以根据实时获取到的算力服务信息和算力路由信息，为各算力接入网关规划多条到能够满足算力服务资源需求的算力服务实例节点所连接的算力服务网关的算力路由路径，为各算力服务网关分配该算力服务网关的算力服务SRv6 SID，基于规划确定的每一条算力路由路径，确定该条算力路由路径经过的算力接入网关的第一转发动作策略和算力服务网关的第二转发动作策略，向每一算力接入网关下发该算力接入网关的第一转发动作策略，并向每一算力服务网关下发该算力服务网关的第二转发动作策略Specifically, in the case of calling a single computing power service (referring to calling a single target computing power service), the computing power routing controller can access the computing power for each computing power according to the computing power service information and computing power routing information obtained in real time. The gateway plans a number of computing power routing paths to the computing power service gateway connected to the computing power service instance node that can meet the computing power service resource requirements, and allocates the computing power service gateway's computing power service SRv6 SID for each computing power service gateway. Based on each computing power routing path determined in the plan, determine the first forwarding action strategy of the computing power access gateway and the second forwarding action strategy of the computing power service gateway through which the computing power routing path passes, and connect to each computing power The gateway delivers the first forwarding action policy of the computing power access gateway, and delivers the second forwarding action policy of the computing power service gateway to each computing power service gateway

步骤202、算力接入网关基于用户设备发送的算力业务流量携带的目标算力服务的任播IPv6地址和算力接入网关的第一转发动作策略，将算力业务流量封装对应的SR扩展头之后，将封装对应的SR扩展头的算力业务流量转发至相应的中继路由节点。Step 202: The computing power access gateway encapsulates the corresponding SR based on the anycast IPv6 address of the target computing power service carried by the computing power service traffic sent by the user equipment and the first forwarding action policy of the computing power access gateway. After the header is extended, the computing power service traffic encapsulated with the corresponding SR extension header is forwarded to the corresponding relay routing node.

具体地，算力接入网关接收算力业务流量之后，可以将算力业务流量携带的目标算力服务的任播IPv6地址与自身的第一转发动作策略中的算力服务SRv6 SID表项进行匹配；基于与目标算力服务的任播IPv6地址匹配的算力服务SRv6 SID，确定实际的转发路径，并为该算力业务流量封装对应的SRH(指SR扩展头)，将实际的转发路径的各跳节点的SID封装在对应的SRH中，然后将封装对应的SRH的算力业务流量转发至相应的中继路由节点。Specifically, after the computing power access gateway receives the computing power service traffic, it can perform the anycast IPv6 address of the target computing power service carried by the computing power service traffic with the SRv6 SID entry of the computing power service in its first forwarding action policy. Matching: Based on the computing service SRv6 SID matching the anycast IPv6 address of the target computing service, the actual forwarding path is determined, and the corresponding SRH (referring to the SR extension header) is encapsulated for the computing service traffic, and the actual forwarding path is encapsulated. The SID of each hop node is encapsulated in the corresponding SRH, and then the computing power service traffic encapsulated in the corresponding SRH is forwarded to the corresponding relay routing node.

步骤203、中继路由节点基于中继路由节点的第三转发动作策略和对应的SR扩展头，将接收的封装对应的SR扩展头的算力业务流量进行转发。Step 203: The relay routing node forwards the received computing power service traffic encapsulated with the corresponding SR extension header based on the third forwarding action policy of the relay routing node and the corresponding SR extension header.

具体地，接收到封装对应的SRH的算力业务流量的任一中继路由节点，可以基于自身的第三转发动作策略，根据对应的SRH中下一跳节点的SRv6 SID，将封装对应的SRH的算力业务流量转发至下一跳节点。Specifically, any relay routing node that receives the computing service traffic encapsulating the corresponding SRH can encapsulate the corresponding SRH based on its own third forwarding action policy and the SRv6 SID of the next hop node in the corresponding SRH. The computing power service traffic is forwarded to the next hop node.

步骤204、算力服务网关基于算力服务网关的第二转发动作策略、对应的SR扩展头和算力服务网关的算力服务SRv6 SID，将算力业务流量转发至目标算力服务实例节点。Step 204: The computing power service gateway forwards the computing power service traffic to the target computing power service instance node based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header, and the computing power service SRv6 SID of the computing power service gateway.

具体地，接收到封装对应的SRH的算力业务流量的算力服务网关，可以将对应的SRH中的目的地址，与自身的算力服务SRv6 SID进行匹配；基于自身的第二转发动作策略，基于与目的地址匹配的一个算力服务SRv6 SID，剥离对应的SRH，将剥离对应的SRH后的算力业务流量转发至目标算力服务实例节点。Specifically, the computing power service gateway that receives the computing power service traffic encapsulating the corresponding SRH can match the destination address in the corresponding SRH with its own computing power service SRv6 SID; based on its own second forwarding action policy, Based on a computing power service SRv6 SID matching the destination address, the corresponding SRH is stripped, and the computing power service traffic after stripping the corresponding SRH is forwarded to the target computing power service instance node.

本发明实施例提供的基于SRv6的算力路由方法，基于本发明上述基于SRv6的算力路由系统，其实施方式与本发明提供的基于SRv6的算力路由系统的实施方式一致，且可以达到相同的有益效果，此处不再赘述。The SRv6-based computing power routing method provided by the embodiment of the present invention is based on the above-mentioned SRv6-based computing power routing system of the present invention, and its implementation is consistent with the SRv6-based computing power routing system provided by the present invention, and can achieve the same The beneficial effects will not be repeated here.

该基于SRv6的算力路由方法，基于前述各实施例的基于SRv6的算力路由系统。因此，在前述各实施例中的基于SRv6的算力路由系统中的描述和定义，可以用于本发明实施例中各方法步骤的理解。The SRv6-based computing power routing method is based on the SRv6-based computing power routing system of the foregoing embodiments. Therefore, the descriptions and definitions in the SRv6-based computing power routing system in the foregoing embodiments can be used to understand the method steps in the embodiments of the present invention.

本发明实施例基于算力服务的任播IP地址和算力服务网关的算力服务SRv6 SID进行算力业务流量的转发，对算力网络的基础设施的改动更小，能够在对算力网络的基础设施较小改动的前提下，实现算力网络资源更高效的调用，能通过算力路由控制器实现了对单算力服务的高效调用，即“以网取算”。The embodiment of the present invention forwards the computing power service traffic based on the anycast IP address of the computing power service and the computing power service SRv6 SID of the computing power service gateway, and the changes to the infrastructure of the computing power network are smaller. Under the premise of minor changes to the infrastructure, more efficient invocation of computing power network resources can be achieved, and efficient invocation of single computing power services can be achieved through the computing power routing controller, that is, "taking computing through the network".

图3是本发明提供的基于SRv6的算力路由方法的流程示意图之二。如图3所示，基于上述任一实施例的内容，基于SRv6的算力路由方法，还包括：步骤301、步骤302、步骤303、步骤304、步骤305和步骤306。FIG. 3 is the second schematic flowchart of the SRv6-based computing power routing method provided by the present invention. As shown in FIG. 3 , based on the content of any of the above embodiments, the SRv6-based computing power routing method further includes:step 301 , step 302 , step 303 , step 304 , step 305 and step 306 .

步骤301、在调用包括多个目标算力服务的算力服务功能链的情况下，算力路由控制器基于算力服务信息、算力路由信息和目标算力服务的调用顺序，确定各算力服务网关的算力服务SRv6 SID、多条算力路由路径及各算力接入网关的第一转发动作策略和各算力服务网关的第二转发动作策略；向每一算力接入网关下发每一算力接入网关的第一转发动作策略；向每一算力服务网关下发每一算力服务网关的算力服务SRv6 SID和第二转发动作策略。Step 301: In the case of invoking a computing power service function chain including multiple target computing power services, the computing power routing controller determines each computing power based on the computing power service information, computing power routing information and the calling sequence of the target computing power service. The computing power service SRv6 SID of the service gateway, multiple computing power routing paths, and the first forwarding action policy of each computing power access gateway and the second forwarding action policy of each computing power service gateway; Send the first forwarding action policy of each computing power access gateway; and deliver the computing power service SRv6 SID and the second forwarding action policy of each computing power service gateway to each computing power service gateway.

具体地，在面向算力服务功能链调用(指调用包括多个目标算力服务的算力服务功能链)的情况下，可以在前述调用单个目标算力服务的基础上，基于SRv6的SFC(业务链，Service Function Chaining)特性，实现用户对由多个算力原子服务(即目标算力服务)构成的算力服务功能链的调用。Specifically, in the case of calling the computing power service function chain oriented (referring to calling the computing power service function chain including multiple target computing power services), on the basis of calling a single target computing power service as described above, the SRv6-based SFC ( Business chain, Service Function Chaining) feature, realizes the user's invocation of the computing power service function chain composed of multiple computing power atomic services (ie target computing power services).

算力服务功能链，可以由多个算力原子服务组成，通过算力原子服务间的按序调用形成一条算力服务功能链。The computing power service function chain can be composed of multiple computing power atomic services, and a computing power service function chain is formed through the sequential invocation of computing power atomic services.

算力路由控制器可以根据获取到的实时的算力服务信息和算力路由信息并根据算力服务功能链所含算力原子服务的顺序，为各算力接入网关规划多条到能够满足算力服务资源需求的算力原子服务实例所连接的算力服务网关设备的服务功能链路由(即算力路由路径)。The computing power routing controller can plan multiple computing power access gateways for each computing power access gateway according to the obtained real-time computing power service information and computing power routing information and according to the order of computing power atomic services contained in the computing power service function chain. The service function link routing (ie, the computing power routing path) of the computing power service gateway device connected to the computing power atomic service instance of the computing power service resource requirement.

算力路由控制器分配算力服务网关的算力服务SRv6 SID，确定各算力接入网关的第一转发动作策略并下发，以及确定各算力服务网关的第一转发动作策略并下发的步骤，与前述调用单个目标算力服务的情况类似，此处不再赘述。The computing power routing controller allocates the computing power service SRv6 SID of the computing power service gateway, determines and issues the first forwarding action policy of each computing power access gateway, and determines and delivers the first forwarding action policy of each computing power service gateway The steps of invoking a single target computing power service are similar to those described above, and will not be repeated here.

步骤302、算力接入网关基于接收的算力业务流量携带的第一个目标算力服务的任播IPv6地址和算力接入网关的第一转发动作策略，将算力业务流量封装对应的SR扩展头之后，将封装对应的SR扩展头的算力业务流量转发至相应的中继路由节点。Step 302: The computing power access gateway encapsulates the corresponding computing power service traffic based on the anycast IPv6 address of the first target computing power service carried by the received computing power service traffic and the first forwarding action policy of the computing power access gateway. After the SR extension header, the computing power service traffic encapsulated with the corresponding SR extension header is forwarded to the corresponding relay routing node.

具体地，用户终端设备在调用算力服务功能链时，向该算力服务功能链的第一个算力原子服务发送算力业务流量，算力业务流量的目的地址配置为第一个算力原子服务的任播IPv6地址。Specifically, when the user terminal device calls the computing power service function chain, it sends the computing power business traffic to the first computing power atomic service of the computing power service function chain, and the destination address of the computing power business traffic is configured as the first computing power The anycast IPv6 address of the atomic service.

算力接入网关封装SRH和转发接收到的算力业务流量的步骤，与前述调用单个目标算力服务的情况类似，此处不再赘述。The steps of encapsulating the SRH and forwarding the received computing service traffic by the computing power access gateway are similar to the aforementioned case of invoking a single target computing power service, and will not be repeated here.

步骤303、中继路由节点基于中继路由节点的第三转发动作策略和对应的SR扩展头，将接收的封装对应的SR扩展头的算力业务流量进行转发。Step 303: The relay routing node forwards the received computing power service traffic encapsulated with the corresponding SR extension header based on the third forwarding action policy of the relay routing node and the corresponding SR extension header.

具体地，中继路由节点转发接收到的算力业务流量的步骤，与前述调用单个目标算力服务的情况类似，此处不再赘述。Specifically, the step of forwarding the received computing power service traffic by the relay routing node is similar to the aforementioned case of invoking a single target computing power service, and will not be repeated here.

步骤304、算力服务网关基于算力服务网关的第二转发动作策略、对应的SR扩展头和算力服务网关的算力服务SRv6 SID，将算力业务流量转发至目标算力服务实例节点。Step 304: The computing power service gateway forwards the computing power service traffic to the target computing power service instance node based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header, and the computing power service SRv6 SID of the computing power service gateway.

具体地，算力服务网关转发接收到的算力业务流量的步骤，与前述调用单个目标算力服务的情况类似，此处不再赘述。Specifically, the step of forwarding the received computing power service traffic by the computing power service gateway is similar to the aforementioned case of invoking a single target computing power service, and will not be repeated here.

步骤305、目标算力服务实例节点更新所述算力业务流量，并更新算力业务流量的目的地址为下一目标算力服务的任播IPv6地址。Step 305: The target computing service instance node updates the computing service traffic, and updates the destination address of the computing service traffic to be the anycast IPv6 address of the next target computing service.

具体地，目标算力服务实例节点完成对算力业务流量的处理后，可以将处理后的算力业务流量返回向其转发算力业务流量的算力服务网关，更新算力业务流量的目的地址为下一目标算力服务的任播IPv6地址。Specifically, after the target computing service instance node completes the processing of the computing service traffic, it can return the processed computing service traffic to the computing service gateway that forwards the computing service traffic to it, and update the destination address of the computing service traffic Anycast IPv6 address serving the next target computing power.

步骤306、算力服务网关对目标算力服务实例节点返回的更新后的算力业务流量封装对应的SR扩展头，基于算力服务网关的第二转发动作策略和对应的SR扩展头，对封装对应的SR扩展头的算力业务流量进行转发。Step 306: The computing power service gateway encapsulates the updated computing power service flow returned by the target computing power service instance node with the corresponding SR extension header, and based on the second forwarding action policy of the computing power service gateway and the corresponding SR extension header, encapsulates the corresponding SR extension header. The computing service traffic of the corresponding SR extension header is forwarded.

具体地，接收到目标算力服务实例节点返回的更新后的算力业务流量的算力服务网关，将算力业务流量的目的地址为下一目标算力服务的任播IPv6地址，并对算力业务流量封装之前剥离的对应的SRH，可以通过非算力服务的任播IP地址的其他公网IP地址，下一目标算力服务的算力服务实例节点发送算力业务流量，从而实现链式调用各目标算力服务。Specifically, the computing power service gateway that receives the updated computing power service flow returned by the target computing power service instance node sets the destination address of the computing power service flow as the anycast IPv6 address of the next target computing power service, and compares the computing power to the computing power service. The corresponding SRH stripped before the encapsulation of the power service traffic can send the computing power service traffic through other public network IP addresses other than the anycast IP address of the computing power service, and the computing power service instance node of the next target computing power service, so as to realize the chain Call each target computing power service.

本发明实施例基于算力服务的任播IP地址和算力服务网关的算力服务SRv6 SID进行算力业务流量的转发，对算力网络的基础设施的改动更小，能够在对算力网络的基础设施较小改动的前提下，实现算力网络资源更高效的调用，能通过算力路由控制器实现了对算力服务功能链的高效调用，即“以网取算”。The embodiment of the present invention forwards the computing power service traffic based on the anycast IP address of the computing power service and the computing power service SRv6 SID of the computing power service gateway, and the changes to the infrastructure of the computing power network are smaller. On the premise of minor changes to the infrastructure of the computing power, more efficient invocation of computing power network resources can be realized, and the efficient invocation of the computing power service function chain can be realized through the computing power routing controller, that is, "taking calculation through the network".

以上所描述的系统实施例仅仅是示意性的，其中作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性的劳动的情况下，即可以理解并实施。The system embodiments described above are only illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place , or distributed to multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到各实施方式可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件。基于这样的理解，上述技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品可以存储在计算机可读存储介质中，如ROM/RAM、磁碟、光盘等，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行各个实施例或者实施例的某些部分的方法。From the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on this understanding, the above-mentioned technical solutions can be embodied in the form of software products in essence or the parts that make contributions to the prior art, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic Disks, optical discs, etc., include instructions for causing a computer device (which may be a personal computer, server, or network device, etc.) to perform the methods of various embodiments or portions of embodiments.

最后应说明的是：以上实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的精神和范围。Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be The technical solutions described in the foregoing embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

Translated fromChinese

1.一种基于SRv6的算力路由系统，其特征在于，包括：算力路由控制器、至少一个算力接入网关、至少一个中继路由节点和至少一个算力服务网关；1. A computing power routing system based on SRv6, comprising: a computing power routing controller, at least one computing power access gateway, at least one relay routing node and at least one computing power service gateway;所述算力路由控制器，用于基于算力服务信息和算力路由信息，确定各所述算力服务网关的算力服务SRv6 SID、多条算力路由路径及各所述算力接入网关的第一转发动作策略和各所述算力服务网关的第二转发动作策略；向每一算力接入网关下发所述每一算力接入网关的第一转发动作策略；向每一算力服务网关下发所述每一算力服务网关的算力服务SRv6 SID和第二转发动作策略；The computing power routing controller is configured to determine, based on the computing power service information and the computing power routing information, the computing power service SRv6 SID of each of the computing power service gateways, multiple computing power routing paths, and each computing power access the first forwarding action policy of the gateway and the second forwarding action policy of each of the computing power service gateways; delivering the first forwarding action policy of each computing power access gateway to each computing power access gateway; A computing power service gateway delivers the computing power service SRv6 SID and the second forwarding action policy of each computing power service gateway;所述算力接入网关，用于基于用户设备发送的算力业务流量携带的目标算力服务的任播IPv6地址和所述算力接入网关的所述第一转发动作策略，将所述算力业务流量封装对应的SR扩展头之后，将所述算力业务流量转发至相应的中继路由节点；The computing power access gateway is configured to, based on the anycast IPv6 address of the target computing power service carried by the computing power service traffic sent by the user equipment and the first forwarding action policy of the computing power access gateway, convert the computing power access gateway. After the computing power service traffic is encapsulated with the corresponding SR extension header, the computing power service traffic is forwarded to the corresponding relay routing node;所述中继路由节点，用于基于所述中继路由节点的第三转发动作策略和所述对应的SR扩展头，将接收的封装所述对应的SR扩展头的所述算力业务流量进行转发；The relay routing node is configured to, based on the third forwarding action policy of the relay routing node and the corresponding SR extension header, perform the received computing power service traffic encapsulating the corresponding SR extension header. Forward;所述算力服务网关，用于基于所述算力服务网关的第二转发动作策略、所述对应的SR扩展头和所述算力服务网关的算力服务SRv6 SID，将所述算力业务流量转发至目标算力服务实例节点；The computing power service gateway is configured to convert the computing power service based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header, and the computing power service SRv6 SID of the computing power service gateway. The traffic is forwarded to the target computing service instance node;其中，所述算力服务信息，包括各算力服务实例节点的负载状态信息、服务能力信息、提供的算力服务的算力服务ID和任播IP地址及连接的所述算力服务网关的IP地址；所述算力路由信息，包括各所述算力接入网关、各所述中继路由节点和各所述算力服务网关的拓扑信息、链路带宽信息、链路时延信息及链路丢包率信息；所述算力路由路径，指从一个所述算力接入网关，经至少一个所述中继路由节点到一个所述算力服务网关的路径；所述目标算力服务的任播IPv6地址，是基于网络服务提供商分配给所述目标算力服务的算力服务提供商的第一信息、所述目标算力服务的ID和补充参数确定的；所述算力服务网关的算力服务SRv6 SID，是基于所述网络服务提供商分配给所述算力服务网关的第二信息、与所述算力服务网关通信连接的各算力服务实例节点提供的算力服务的ID和补充参数确定的；所述目标算力服务实例节点，包括至少一个与所述算力服务网关通信连接的算力服务实例节点。The computing power service information includes the load status information, service capability information of each computing power service instance node, the computing power service ID and anycast IP address of the computing power service provided, and the connected computing power service gateway. IP address; the computing power routing information, including the topology information, link bandwidth information, link delay information and Link packet loss rate information; the computing power routing path refers to the path from one of the computing power access gateways to one of the computing power service gateways via at least one of the relay routing nodes; the target computing power The anycast IPv6 address of the service is determined based on the first information of the computing power service provider assigned by the network service provider to the target computing power service, the ID of the target computing power service, and supplementary parameters; the computing power The computing power service SRv6 SID of the service gateway is based on the second information allocated to the computing power service gateway by the network service provider, and the computing power provided by each computing power service instance node communicatively connected to the computing power service gateway. The ID of the service and the supplementary parameters are determined; the target computing power service instance node includes at least one computing power service instance node that is communicatively connected to the computing power service gateway.2.根据权利要求1所述的基于SRv6的算力路由系统，其特征在于，所述算力路由控制器，还用于基于所述算力服务信息和所述算力路由信息，确定每一条所述算力路由路径的优先级和/或权重；2. The computing power routing system based on SRv6 according to claim 1, wherein the computing power routing controller is further configured to determine each route based on the computing power service information and the computing power routing information. the priority and/or weight of the computing power routing path;所述算力接入网关的所述第一转发动作策略，包括经过所述算力接入网关的所述算力路由路径的优先级和/或权重。The first forwarding action policy of the computing power access gateway includes the priority and/or weight of the computing power routing path passing through the computing power access gateway.3.根据权利要求1或2所述的基于SRv6的算力路由系统，其特征在于，所述算力路由控制器，还用于基于新的算力服务信息和新的算力路由信息，确定上一次下发的各第一转发动作策略和各第二转发动作策略是否满足需求；在不满足需求的情况下，基于所述新的算力服务信息和所述新的算力路由信息，重新确定多条算力路由路径及各算力接入网关的第一转发动作策略和各算力服务网关的转发动作策略；向每一算力接入网关下发重新确定的所述每一算力接入网关的转发动作策略；向每一算力服务网关下发重新确定的所述每一算力服务网关的第二转发动作策略。3. The SRv6-based computing power routing system according to claim 1 or 2, wherein the computing power routing controller is also used to determine based on new computing power service information and new computing power routing information. Whether each first forwarding action policy and each second forwarding action policy issued last time meet the requirements; if the requirements are not met, based on the new computing power service information and the new computing power routing information, Determine multiple computing power routing paths and the first forwarding action policy of each computing power access gateway and the forwarding action policy of each computing power service gateway; deliver the re-determined each computing power to each computing power access gateway forwarding action policy of the access gateway; delivering the re-determined second forwarding action policy of each computing power serving gateway to each computing power serving gateway.4.根据权利要求1所述的基于SRv6的算力路由系统，其特征在于，所述算力路由控制器，还用于基于算力服务信息和算力路由信息，确定各所述中继路由节点的第三转发动作策略；向每一中继路由节点下发所述每一中继路由节点的第三转发动作策略。4. The SRv6-based computing power routing system according to claim 1, wherein the computing power routing controller is further configured to determine each relay route based on computing power service information and computing power routing information the third forwarding action policy of the node; delivering the third forwarding action policy of each relay routing node to each relay routing node.5.根据权利要求1所述的基于SRv6的算力路由系统，其特征在于，所述算力路由控制器，还用于从算力服务编排器，获取所述算力服务信息。5 . The computing power routing system based on SRv6 according to claim 1 , wherein the computing power routing controller is further configured to obtain the computing power service information from a computing power service orchestrator. 6 .6.根据权利要求2所述的基于SRv6的算力路由系统，其特征在于，各所述算力接入网关、各所述中继路由节点和各所述算力服务网关，还用于维护链路状态测量会话，以获取所述链路时延信息及链路丢包率信息；6. The SRv6-based computing power routing system according to claim 2, wherein each of the computing power access gateways, each of the relay routing nodes and each of the computing power service gateways is also used for maintenance a link state measurement session to obtain the link delay information and link packet loss rate information;所述算力路由控制器，还用于从各所述算力接入网关、各所述中继路由节点和各所述算力服务网关，获取所述算力路由信息。The computing power routing controller is further configured to acquire the computing power routing information from each of the computing power access gateways, each of the relay routing nodes, and each of the computing power service gateways.7.根据权利要求6所述的基于SRv6的算力路由系统，其特征在于，所述算力接入网关，还用于基于获取到的所述链路时延信息及链路丢包率信息，更新所述算力接入网关的所述第一转发动作策略中的所述算力路由路径的优先级和/或权重。7 . The computing power routing system based on SRv6 according to claim 6 , wherein the computing power access gateway is further configured to be based on the acquired link delay information and link packet loss rate information. 8 . , updating the priority and/or weight of the computing power routing path in the first forwarding action policy of the computing power access gateway.8.根据权利要求6所述的基于SRv6的算力路由系统，其特征在于，所述链路状态测量会话基于的协议包括Internet控制报文协议或双向主动测量协议。8 . The computing power routing system based on SRv6 according to claim 6 , wherein the protocol on which the link state measurement session is based comprises Internet Control Message Protocol or Bidirectional Active Measurement Protocol. 9 .9.一种基于如权利要求1至8任一所述的基于SRv6的算力路由系统的算力路由方法，其特征在于，包括：9. A computing power routing method based on the computing power routing system based on SRv6 as described in any one of claims 1 to 8, characterized in that, comprising:在调用单个目标算力服务的情况下，算力路由控制器基于算力服务信息和算力路由信息，确定各算力服务网关的算力服务SRv6 SID、多条算力路由路径及各算力接入网关的第一转发动作策略和各所述算力服务网关的第二转发动作策略；向每一算力接入网关下发所述每一算力接入网关的第一转发动作策略；向每一算力服务网关下发所述每一算力服务网关的算力服务SRv6 SID和第二转发动作策略；In the case of invoking a single target computing power service, the computing power routing controller determines the computing power service SRv6 SID, multiple computing power routing paths and each computing power of each computing power service gateway based on the computing power service information and computing power routing information. a first forwarding action policy of the access gateway and a second forwarding action policy of each of the computing power service gateways; delivering the first forwarding action policy of each computing power access gateway to each computing power access gateway; delivering the computing power service SRv6 SID and the second forwarding action policy of each computing power service gateway to each computing power service gateway;所述算力接入网关基于用户设备发送的算力业务流量携带的目标算力服务的任播IPv6地址和所述算力接入网关的所述第一转发动作策略，将所述算力业务流量封装对应的SR扩展头之后，将封装所述对应的SR扩展头的所述算力业务流量转发至相应的中继路由节点；The computing power access gateway transfers the computing power service based on the anycast IPv6 address of the target computing power service carried by the computing power service traffic sent by the user equipment and the first forwarding action policy of the computing power access gateway. After the traffic encapsulates the corresponding SR extension header, forward the computing power service traffic encapsulated with the corresponding SR extension header to the corresponding relay routing node;中继路由节点基于所述中继路由节点的第三转发动作策略和所述对应的SR扩展头，将接收的封装所述对应的SR扩展头的所述算力业务流量进行转发；The relay routing node forwards the received computing power service traffic encapsulating the corresponding SR extension header based on the third forwarding action policy of the relay routing node and the corresponding SR extension header;所述算力服务网关基于所述算力服务网关的第二转发动作策略、所述对应的SR扩展头和所述算力服务网关的算力服务SRv6 SID，将所述算力业务流量转发至目标算力服务实例节点。The computing power service gateway forwards the computing power service traffic to the computing power service gateway based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header and the computing power service The target computing service instance node.10.根据权利要求9所述的基于SRv6的算力路由方法，其特征在于，还包括：10. The SRv6-based computing power routing method according to claim 9, further comprising:在调用包括多个所述目标算力服务的算力服务功能链的情况下，算力路由控制器基于所述算力服务信息、所述算力路由信息和所述目标算力服务的调用顺序，确定各算力服务网关的算力服务SRv6 SID、多条算力路由路径及各算力接入网关的第一转发动作策略和各所述算力服务网关的第二转发动作策略；向每一算力接入网关下发所述每一算力接入网关的第一转发动作策略；向每一算力服务网关下发所述每一算力服务网关的算力服务SRv6SID和第二转发动作策略；In the case of invoking a computing power service function chain including a plurality of the target computing power services, the computing power routing controller is based on the computing power service information, the computing power routing information and the calling sequence of the target computing power service. , determine the computing power service SRv6 SID of each computing power service gateway, multiple computing power routing paths, and the first forwarding action strategy of each computing power access gateway and the second forwarding action strategy of each computing power service gateway; A computing power access gateway delivers the first forwarding action policy of each computing power access gateway; delivers the computing power service SRv6SID and second forwarding of each computing power service gateway to each computing power service gateway action strategy;所述算力接入网关基于接收的算力业务流量携带的第一个目标算力服务的任播IPv6地址和所述算力接入网关的所述第一转发动作策略，将所述算力业务流量封装对应的SR扩展头之后，将封装所述对应的SR扩展头的所述算力业务流量转发至相应的中继路由节点；The computing power access gateway transfers the computing power based on the anycast IPv6 address of the first target computing power service carried by the received computing power service traffic and the first forwarding action policy of the computing power access gateway. After the service traffic encapsulates the corresponding SR extension header, forward the computing power service traffic encapsulated with the corresponding SR extension header to the corresponding relay routing node;所述中继路由节点基于所述中继路由节点的第三转发动作策略和所述对应的SR扩展头，将接收的封装所述对应的SR扩展头的所述算力业务流量进行转发；The relay routing node forwards the received computing power service traffic encapsulating the corresponding SR extension header based on the third forwarding action policy of the relay routing node and the corresponding SR extension header;所述算力服务网关基于所述算力服务网关的第二转发动作策略、所述对应的SR扩展头和所述算力服务网关的算力服务SRv6 SID，将所述算力业务流量转发至目标算力服务实例节点；The computing power service gateway forwards the computing power service traffic to the computing power service gateway based on the second forwarding action policy of the computing power service gateway, the corresponding SR extension header and the computing power service The target computing service instance node;所述目标算力服务实例节点更新所述算力业务流量，并更新所述算力业务流量的目的地址为下一所述目标算力服务的任播IPv6地址；The target computing power service instance node updates the computing power service flow, and updates the destination address of the computing power service flow to be the anycast IPv6 address of the next target computing power service;所述算力服务网关对所述目标算力服务实例节点返回的更新后的所述算力业务流量封装所述对应的SR扩展头，基于所述算力服务网关的第二转发动作策略和所述对应的SR扩展头，对封装所述对应的SR扩展头的所述算力业务流量进行转发。The computing power service gateway encapsulates the corresponding SR extension header for the updated computing power service flow returned by the target computing power service instance node, based on the second forwarding action policy of the computing power service gateway and the The corresponding SR extension header is forwarded, and the computing power service traffic encapsulated with the corresponding SR extension header is forwarded.

Images