WO2023273940A1

WO2023273940A1 - Virtual network optimization method and apparatus, and computer storage medium

Info

Publication number: WO2023273940A1
Application number: PCT/CN2022/099902
Authority: WO
Inventors: 王大江; 陈然; 王其磊
Original assignee: 中兴通讯股份有限公司
Priority date: 2021-06-28
Filing date: 2022-06-20
Publication date: 2023-01-05
Also published as: CN115603850A

Abstract

A virtual network optimization method and apparatus, and a computer storage medium. The virtual network optimization method comprises: generating at least two candidate optimization solutions on the basis of a current user intention; performing tradeoff analysis on the candidate optimization solutions, and determining a target optimization solution from candidate optimization solutions satisfying a tradeoff analysis threshold requirement; determining, according to a traffic change rule of a physical network, an optimization time window for executing the target optimization solution; and re-establishing a mapping relationship according to the target optimization solution and the optimization time window.

Description

虚拟网络的优化方法、装置及计算机存储介质Virtual network optimization method, device and computer storage medium

相关申请的交叉引用Cross References to Related Applications

本申请基于申请号为202110719849.4，申请日为2021年06月28日的中国专利申请提出，并要求该中国专利申请的优先权，该中国专利申请的全部内容在此引入本申请作为参考。This application is based on a Chinese patent application with application number 202110719849.4 and a filing date of June 28, 2021, and claims the priority of this Chinese patent application. The entire content of this Chinese patent application is hereby incorporated by reference into this application.

技术领域technical field

本申请涉及网络优化技术领域，尤其涉及一种虚拟网络的优化方法、装置及计算机存储介质。The present application relates to the technical field of network optimization, in particular to a virtual network optimization method, device and computer storage medium.

背景技术Background technique

在数字化转型时代，OTN(Optical Transport Network，光传送网)切片/OVPN(Optical Virtual Private Network，光虚拟私有网)专网是OTN网络在政企、金融、云网智联等应用场景中提供的一种极为重要的OTN网络服务技术，具有服务定制化、服务协议等级明确、能够精细管控和扩大盈利等特性，可以给不同的网络用户分配不同的网络资源，实现资源的预分配和预优化，对业务的带宽、时延等进行精确控制，以实现对网络资源的充分和有效利用。In the era of digital transformation, OTN (Optical Transport Network, Optical Transport Network) slicing/OVPN (Optical Virtual Private Network, Optical Virtual Private Network) private network is an OTN network provided in application scenarios such as government and enterprise, finance, and cloud-network intelligence. An extremely important OTN network service technology, which has the characteristics of customized services, clear service agreement levels, fine control, and profit expansion. It can allocate different network resources to different network users and realize resource pre-allocation and pre-optimization. Precisely control the bandwidth and delay of services to achieve full and effective use of network resources.

意图网络的出现迎合了OTN切片/OVPN专网的运行特点和对业务的敏捷性要求，意图网络技术通过构建虚拟网络和物理网络之间的映射关系，实现基于用户意图驱动的网络资源灵活分配。但是基于意图网络技术如何对OTN切片/OVPN专网等虚拟网络进行优化和校正，目前业内还没提出相应的解决方案。The emergence of the intent network caters to the operating characteristics of the OTN slice/OVPN private network and the agility requirements for services. The intent network technology realizes the flexible allocation of network resources driven by user intent by building a mapping relationship between the virtual network and the physical network. However, how to optimize and correct virtual networks such as OTN slices/OVPN private networks based on intent-based network technology has not yet proposed a corresponding solution in the industry.

发明内容Contents of the invention

以下是对本文详细描述的主题的概述。本概述并非是为了限制权利要求的保护范围。The following is an overview of the topics described in detail in this article. This summary is not intended to limit the scope of the claims.

本申请实施例提供了一种虚拟网络的优化方法、装置及计算机存储介质。Embodiments of the present application provide a virtual network optimization method, device, and computer storage medium.

第一方面，本申请实施例提供了一种虚拟网络的优化方法，所述虚拟网络的虚拟链路与物理网络的物理网络资源构成映射关系，所述优化方法包括：在所述映射关系不满足当前用户意图的情况下，基于当前用户意图生成至少两个候选优化方案；对所述候选优化方案进行权衡分析，并从符合权衡分析门限要求的候选优化方案中确定目标优化方案；根据所述物理网络的流量变化规律确定执行所述目标优化方案的优化时间窗口；根据所述目标优化方案和所述优化时间窗口重新建立映射关系。In the first aspect, the embodiment of the present application provides a method for optimizing a virtual network. The virtual links of the virtual network and the physical network resources of the physical network form a mapping relationship. The optimization method includes: when the mapping relationship does not satisfy In the case of the current user intention, generate at least two candidate optimization schemes based on the current user intention; perform a trade-off analysis on the candidate optimization schemes, and determine the target optimization scheme from the candidate optimization schemes that meet the trade-off analysis threshold requirements; according to the physical An optimization time window for executing the target optimization scheme is determined according to the flow change law of the network; and a mapping relationship is re-established according to the target optimization scheme and the optimization time window.

第二方面，本申请实施例提供了一种虚拟网络的优化装置，包括至少一个处理器和用于与所述至少一个处理器通信连接的存储器；所述存储器存储有能够被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如第一方面所述的虚拟网络的优化方法。In the second aspect, the embodiment of the present application provides a device for optimizing a virtual network, including at least one processor and a memory for communicating with the at least one processor; the memory stores information that can be processed by the at least one processor. An instruction executed by a processor, the instruction is executed by the at least one processor, so that the at least one processor can execute the method for optimizing a virtual network as described in the first aspect.

第三方面，本申请实施例还提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机可执行指令，所述计算机可执行指令用于使计算机执行如第一方面所述的虚拟网络的优化方法。In the third aspect, the embodiment of the present application also provides a computer-readable storage medium, the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to make the computer execute the The optimization method of the virtual network.

本申请的其它特征和优点将在随后的说明书中阐述，并且，部分地从说明书中变得显而易见，或者通过实施本申请而了解。本申请的目的和其他优点可通过在说明书、权利要求书以及附图中所特别指出的结构来实现和获得。Additional features and advantages of the application will be set forth in the description which follows, and, in part, will be obvious from the description, or may be learned by practice of the application. The objectives and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

附图说明Description of drawings

附图用来提供对本申请技术方案的进一步理解，并且构成说明书的一部分，与本申请的示例一起用于解释本申请的技术方案，并不构成对本申请技术方案的限制。The accompanying drawings are used to provide a further understanding of the technical solution of the present application, and constitute a part of the description, and are used together with the examples of the application to explain the technical solution of the application, and do not constitute a limitation to the technical solution of the application.

图1是本申请一个实施例提供的虚拟网络的优化方法的整体方法流程图；FIG. 1 is an overall method flowchart of a method for optimizing a virtual network provided by an embodiment of the present application;

图2是本申请一个实施例提供的生成候选优化方案的流程图；FIG. 2 is a flow chart of generating candidate optimization schemes provided by an embodiment of the present application;

图3是本申请一个实施例提供的候选优化方案权衡分析的流程图；FIG. 3 is a flowchart of a trade-off analysis of candidate optimization schemes provided by an embodiment of the present application;

图4是本申请一个实施例提供的评估候选优化方案得到目标优化方案的流程图；FIG. 4 is a flow chart of evaluating candidate optimization schemes to obtain target optimization schemes provided by an embodiment of the present application;

图5是本申请一个实施例提供的计算执行优化方案的时间窗口的流程图；FIG. 5 is a flowchart of a time window for calculating and executing an optimization scheme provided by an embodiment of the present application;

图6是本申请示例提供的虚拟网络和物理网络映射关系意图；Fig. 6 is an illustration of the mapping relationship between the virtual network and the physical network provided by the example of this application;

图7是本申请示例提供的物理连接重优化的物理网络拓扑图；FIG. 7 is a physical network topology diagram of physical connection re-optimization provided by an example of the present application;

图8是本申请示例提供的方案一的物理网络拓扑图；Fig. 8 is a physical network topology diagram of Solution 1 provided by the example of this application;

图9是本申请示例提供的方案二的物理网络拓扑图；FIG. 9 is a physical network topology diagram of Solution 2 provided by the example of this application;

图10是本申请示例提供的流量统计示意图；Figure 10 is a schematic diagram of traffic statistics provided by the example of this application;

图11是本申请实施例提供的虚拟网络的优化装置的结构示意图。FIG. 11 is a schematic structural diagram of an apparatus for optimizing a virtual network provided by an embodiment of the present application.

具体实施方式detailed description

为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, not to limit the present application.

随着数字化转型时代的到来，业务敏捷性几乎已经成为数字化转型的代名词和基本特征。要想实现和提高OTN网络服务的敏捷性，就需要将智能OTN网络管控体系逐步转型到依托网络DT(Digital Twins，数字孪生)仿真与分析技术，集SDN(Software Defined Network，软件定义网络)、AI(Artificial Intelligence，人工智能)技术应用于一体的基于意图驱动的IBON(Intent Based Optical Network，智简光网络)，并在此基础上推动OTN网络服务走向跨完整生命周期的自动化生态体系建设。With the advent of the era of digital transformation, business agility has almost become synonymous and basic characteristics of digital transformation. In order to realize and improve the agility of OTN network services, it is necessary to gradually transform the intelligent OTN network management and control system to rely on network DT (Digital Twins, digital twin) simulation and analysis technology, integrating SDN (Software Defined Network, software defined network), AI (Artificial Intelligence, artificial intelligence) technology is applied to the integrated intent-driven IBON (Intent Based Optical Network, Intelligent and Simple Optical Network), and on this basis, it promotes the construction of an automated ecosystem for OTN network services across the complete life cycle.

在IBON中，意图指的是用户网络服务的想法或方案，用于描述用户想要网络服务所达到的某种状态，意图网络就是将这些方案转化成网络服务编排策略，策略编排无误后就可以下发配置到网络设备，并实时监测网络设备的状态，不断校验和优化意图网络服务编排策略。IBON的这些特点恰恰迎合了数转时代对用户OTN网络服务领悟与认知的敏捷性要求。In IBON, intent refers to the idea or plan of the user's network service, which is used to describe a certain state that the user wants the network service to achieve. The intent network is to transform these plans into a network service orchestration strategy. After the strategy arrangement is correct, it can be Deliver configurations to network devices, monitor the status of network devices in real time, and continuously verify and optimize intent network service orchestration strategies. These characteristics of IBON just cater to the agility requirements of users' OTN network service comprehension and cognition in the era of digital transformation.

基于此，本申请实施例提供了一种虚拟网络的优化方法，装置及计算机存储介质，为基于OTN网络的虚拟网络提供了优化和校正的解决方案，基于意图驱动虚拟网络和物理网络之间映射关系的自动调整，从而满足业务的敏捷性要求。Based on this, the embodiment of the present application provides a virtual network optimization method, device and computer storage medium, which provides an optimization and correction solution for the virtual network based on the OTN network, and drives the mapping between the virtual network and the physical network based on the intent Automatic adjustment of relationships to meet business agility requirements.

参照图1，本申请实施例提供了一种虚拟网络的优化方法，虚拟网络的虚拟链路与物理网络的物理网络资源构成映射关系，优化方法包括但不限于以下步骤S100、步骤S200、步骤 S300和步骤S400。Referring to FIG. 1 , an embodiment of the present application provides a method for optimizing a virtual network. The virtual links of the virtual network and the physical network resources of the physical network form a mapping relationship. The optimization method includes but is not limited to the following steps S100, S200, and S300 and step S400.

步骤S100，在映射关系不满足当前用户意图的情况下，基于当前用户意图生成至少两个候选优化方案。Step S100, if the mapping relationship does not satisfy the current user intention, generate at least two candidate optimization schemes based on the current user intention.

步骤S200，对候选优化方案进行权衡分析，并从符合权衡分析门限要求的候选优化方案中确定目标优化方案。Step S200, performing a tradeoff analysis on candidate optimization schemes, and determining a target optimization scheme from candidate optimization schemes meeting the threshold requirements of the tradeoff analysis.

步骤S300，根据物理网络的流量变化规律确定执行目标优化方案的优化时间窗口。In step S300, an optimization time window for executing the target optimization solution is determined according to the traffic variation rule of the physical network.

步骤S400，根据目标优化方案和优化时间窗口重新建立映射关系。Step S400, re-establish the mapping relationship according to the target optimization scheme and the optimization time window.

本申请实施例的虚拟网络与物理网络存在映射关系，虚拟网络通常是软件定义的网络，例如OTN切片/OVPN专网等，为便于说明，本申请实施例以虚拟网络是OTN切片/OVPN专网为例子进行说明。可以理解的是，本申请实施例中，认为OTN切片与OVPN专网指的是同一种网络服务技术，均适用于ODU(Optical channel Data Unit，光通路数据单元)层、OCH(Optical channel，光通道)层两种情况。上述两种虚拟网络均基于OTN网络服务技术为不同用户的业务需求提供专用网络资源，通过在一个独立的物理网络上切分创建出多个逻辑的OTN切片/OVPN专网，根据OTN切片的SLA(Service Level Agreement，服务级别协议)等级，实现资源的预分配、预优化，对不同OTN切片上业务的带宽、时延等进行精确控制，以实现对网络资源的充分和有效利用。There is a mapping relationship between the virtual network and the physical network in the embodiment of the present application. The virtual network is usually a software-defined network, such as an OTN slice/OVPN private network. Illustrate with an example. It can be understood that in the embodiment of the present application, it is considered that OTN slice and OVPN private network refer to the same network service technology, and both are applicable to ODU (Optical channel Data Unit, optical channel data unit) layer, OCH (Optical channel, optical channel) layer in two cases. Both of the above two virtual networks are based on OTN network service technology to provide dedicated network resources for different users' business needs. Multiple logical OTN slices/OVPN private networks are created by segmenting on an independent physical network. According to the SLA of OTN slices (Service Level Agreement, service level agreement) level, realize resource pre-allocation, pre-optimization, and precisely control the bandwidth and delay of services on different OTN slices, so as to realize the full and effective use of network resources.

随着IBON意图化网络服务理念的出现，OTN切片/OVPN专网与物理网络资源的映射关系也逐步形成意图化趋势，实际上意图化的映射关系，就是用户切分创建意图OTN切片/OVPN专网服务的集中体现：将用户需要的OTN切片/OVPN专网的服务意图转换成OTN切片/OVPN专网服务与物理网络资源的映射策略，以满足用户对专网服务的意图要求。With the emergence of the IBON intentional network service concept, the mapping relationship between OTN slices/OVPN private networks and physical network resources has gradually formed an intentional trend. In fact, the intentional mapping relationship is that users create intentional OTN slices/OVPN Concentrated embodiment of network services: convert the service intention of OTN slice/OVPN private network required by users into a mapping strategy of OTN slice/OVPN private network service and physical network resources, so as to meet the intention requirements of users for private network services.

在实际的IBON意图网络开通和运维过程中，虚拟网络和物理网络之间的映射关系决定了OTN切片/OVPN专网占用的物理网络传输性能，同时用户的业务需求和物理网络的资源情况也并非一成不变，因此除了在创建开通虚拟网络阶段，需要考虑虚拟网络和物理网络之间映射关系的合理性，在虚拟网络的运维阶段，还需要考虑虚拟网络和物理网络之间映射关系的优化和校正，以保证满足虚拟网络和物理网络的一致性要求，这也是基于意图的网络服务的敏捷性的重要体现。In the actual IBON intent network provisioning and O&M process, the mapping relationship between the virtual network and the physical network determines the transmission performance of the physical network occupied by the OTN slice/OVPN private network. It is not static. Therefore, in addition to the stage of creating and opening the virtual network, it is necessary to consider the rationality of the mapping relationship between the virtual network and the physical network. In the operation and maintenance stage of the virtual network, it is also necessary to consider the optimization and Calibration to ensure that the consistency requirements of the virtual network and the physical network are met, which is also an important manifestation of the agility of the intent-based network service.

本申请实施例基于虚拟网络在运维过程中映射关系无法满足当前用户意图的前提下提出，基于当前用户意图重新创建虚拟网络和物理网络之间的映射关系。其中，造成映射关系不满足当前用户意图的情况包括以下至少之一。The embodiment of the present application is proposed based on the premise that the mapping relationship of the virtual network cannot meet the current user's intention during the operation and maintenance process, and recreates the mapping relationship between the virtual network and the physical network based on the current user's intention. Wherein, the situation causing the mapping relationship not satisfying the current user intention includes at least one of the following.

①.用户意图发生改变，表现为OTN切片/OVPN专网可能无法满足当前用户需求，即虚拟网络与物理网络之间的映射关系不满足当前用户意图。①. The user's intention changes, which means that the OTN slice/OVPN private network may not be able to meet the current user's needs, that is, the mapping relationship between the virtual network and the physical network does not meet the current user's intention.

②.OTN切片/OVPN专网扩容、增量规划，需要将OTN切片/OVPN专网中包含的原有虚拟链路与新扩容的虚拟链路统一起来做意图映射，这样原有虚拟链路与物理网络资源的映射关系就要被删除，代之的是对统一后的虚拟链路与物理网络资源的映射关系。②. OTN slicing/OVPN private network expansion and incremental planning, it is necessary to unify the original virtual link contained in the OTN slicing/OVPN private network with the newly expanded virtual link for intent mapping, so that the original virtual link and The mapping relationship of physical network resources will be deleted and replaced by the unified mapping relationship between virtual links and physical network resources.

③.部分映射的物理连接或者链路因为故障而出现倒换或者重路由，但重路由物理连接或者链路数量累积到一定程度，导致OTN切片/OVPN专网包含的所有虚拟链路与物理网络资源的实际映射关系与当前用户意图发生严重背离时，就需要考虑基于意图对OTN切片/OVPN专网与物理网络资源的实际映射关系进行优化与校正。③. Partially mapped physical connections or links are switched or rerouted due to failures, but the number of rerouted physical connections or links accumulates to a certain extent, resulting in all virtual links and physical network resources included in the OTN slice/OVPN private network When there is a serious deviation between the actual mapping relationship of the OTN slice/OVPN private network and the current user intention, it is necessary to consider optimizing and correcting the actual mapping relationship between the OTN slice/OVPN private network and the physical network resources based on the intention.

④.随着物理网络的网元设备、光纤等使用寿命的增加、老化等原因，部分映射的物理连接或者链路出现性能劣化(误码率增加、抖动增加、时延抖动增加)等情况，无法满足OTN切片/OVPN专网的服务要求。④. With the increase of the service life and aging of network element equipment and optical fibers in the physical network, the performance of some mapped physical connections or links is degraded (increased bit error rate, increased jitter, increased delay jitter), etc., It cannot meet the service requirements of OTN slice/OVPN private network.

⑤.采用AI技术预测OTN切片/OVPN专网所映射的物理网络资源在未来某一时刻发生一定规模的网络故障，从而造成原有映射关系的破坏，需要规避预测发生故障的物理网络资源并编排构建新的映射关系，并满足当前用户意图的要求。⑤. Use AI technology to predict that the physical network resources mapped by the OTN slice/OVPN private network will have a certain scale of network failure at a certain point in the future, which will cause damage to the original mapping relationship. It is necessary to avoid predicting the physical network resources that have failed and arrange them Build a new mapping relationship and meet the requirements of the current user intent.

针对上述情况，适时地优化与校正OTN切片/OVPN专网与物理网络资源的映射关系，满足当前用户意图(或者当前意图映射策略要求)等就是IBON体系的关键技术与重要功能。In view of the above situation, it is the key technology and important function of the IBON system to optimize and correct the mapping relationship between OTN slice/OVPN private network and physical network resources in a timely manner to meet the current user intention (or the current intention mapping policy requirements).

但如何实现基于意图对OTN切片/OVPN专网的优化与校正，才能确保因优化与校正操作引起的OTN切片/OVPN专网中部分服务的中断，从时间、空间上对现网运营整体稳定性的影响最小，从而满足网络运营的“求稳”诉求，同时使得优化校正后的OTN切片/OVPN专网能为用户运营带来更多的收益期望、更加卓越的用户体验、更好的网络性能，是在数字化转型时代IBON体系下的OTN智能运维需要解决的关键技术问题。下面将详细对虚拟网络的优化和校正过程进行说明。However, how to realize the optimization and correction of OTN slices/OVPN private networks based on intent can ensure the interruption of some services in OTN slices/OVPN private networks caused by optimization and correction operations, and ensure the overall stability of live network operations in terms of time and space. The impact of the OTN slice/OVPN private network after optimization and correction can be minimized, thereby meeting the "stability" demands of network operations, and at the same time, the optimized and corrected OTN slice/OVPN private network can bring more revenue expectations, more excellent user experience, and better network performance to user operations , is a key technical issue that needs to be solved for OTN intelligent O&M under the IBON system in the era of digital transformation. The optimization and correction process of the virtual network will be described in detail below.

针对上述步骤S100中基于用户意图生成的多个候选优化方案，对其进行权衡分析。当前网络运行环境下，并非所有优化方案都满足电信运营商的要求；为保障电信运营商的网络服务的长期稳定运行，本步骤基于一定的门限要求对候选优化方案进行筛选，丢弃不符合电信运营商服务要求的候选优化方案，进而确定满足电信运营商服务要求的候选优化方案，以满足电信运营商的风险管控、收益增效等要求。符合权衡分析门限要求的这些候选优化方案将进行下一步的筛选，本申请实施例提供了一套对候选优化方案进行评估的评估方案，基于该评估方案得到最优结果的候选优化方案，将作为用于执行优化和校正虚拟网络的目标优化方案，当然，除了采用最优结果作为目标优化方案，还可以采用次优结果作为目标优化方案，这个可以根据实际网络服务要求确定。具体的权衡分析方法和目标优化方案确定方法，将在后面进行详细说明。A trade-off analysis is performed on multiple candidate optimization schemes generated based on user intentions in the above step S100. Under the current network operating environment, not all optimization schemes meet the requirements of telecom operators; in order to ensure the long-term stable operation of telecom operators’ network services, this step screens candidate optimization schemes based on certain threshold requirements, and discards those that do not meet the requirements of telecom operators. Candidate optimization schemes that meet the service requirements of telecom operators, and then determine the candidate optimization schemes that meet the service requirements of telecom operators, so as to meet the requirements of telecom operators for risk control and revenue enhancement. These candidate optimization schemes that meet the threshold requirements of the trade-off analysis will be screened in the next step. The embodiment of the present application provides a set of evaluation schemes for evaluating candidate optimization schemes. The candidate optimization schemes that obtain the best results based on the evaluation schemes will be used as The target optimization scheme used to optimize and correct the virtual network, of course, in addition to using the optimal result as the target optimization solution, the suboptimal result can also be used as the target optimization solution, which can be determined according to the actual network service requirements. The specific trade-off analysis method and the method for determining the target optimization scheme will be described in detail later.

众所周知，物理网络在不同时期具有不同的流量特性，例如傍晚的时候达到流量最高峰，在深夜的时候达到流量最低谷等，在不同时间段执行上述目标优化方案，对实际网络将造成不同程度的影响。因此，尽管通过步骤S200可以得到目标优化方案，但该方案仅表示优化后的情况，并没有考虑优化过程中带来的业务中断等问题。因此步骤S300基于物理网络的流量变化规律计算出执行优化方案的优化时间窗口，例如选择流量最小的时间窗口执行该优化方案，使得优化过程对物理网络造成的影响降低到最小。As we all know, the physical network has different traffic characteristics in different periods, such as reaching the highest traffic peak in the evening and reaching the lowest traffic valley in the middle of the night, etc. Executing the above target optimization scheme in different time periods will cause different degrees of impact on the actual network influences. Therefore, although the target optimization solution can be obtained through step S200, the solution only represents the optimized situation, and does not take into account problems such as service interruption caused by the optimization process. Therefore, step S300 calculates the optimization time window for executing the optimization scheme based on the traffic variation law of the physical network, for example, selects the time window with the smallest traffic to execute the optimization scheme, so that the impact of the optimization process on the physical network is minimized.

通过上述步骤S100至步骤S400，可以基于意图重新构建虚拟网络和物理网络之间的映射关系。其中参照图2，步骤S100中生成候选优化方案的具体方法可以按照以下的步骤S110、步骤S120、步骤S130和步骤S140实现。Through the above steps S100 to S400, the mapping relationship between the virtual network and the physical network can be reconstructed based on intent. Referring to FIG. 2 , the specific method for generating candidate optimization schemes in step S100 can be implemented in accordance with the following steps S110 , S120 , S130 and S140 .

步骤S110，获取当前用户意图、虚拟网络的业务分布信息和物理网络的拓扑信息。Step S110, acquiring the current user intention, service distribution information of the virtual network, and topology information of the physical network.

步骤S120，根据用户意图确定意图指标和意图指标的权重。Step S120, determining the intention index and the weight of the intention index according to the user intention.

步骤S130，根据意图指标和意图指标的权重构建理想目标函数。Step S130, constructing an ideal objective function according to the intention index and the weight of the intention index.

步骤S140，基于理想目标函数和业务分布信息、拓扑信息，通过意图策略编排得到至少两个候选优化方案。Step S140, based on the ideal objective function, service distribution information, and topology information, at least two candidate optimization schemes are obtained through intent policy orchestration.

可以理解的是，无论是由于哪些因素引起映射关系不满足当前用户意图，在重新创建映射关系的时候，都要考虑当前用户意图这一因素，如果生成候选优化方案的时候，用户意图没有改变，则将其作为当前用户意图，如果生成候选优化方案的时候，用户意图发生改变，则将改变后的用户意图作为当前用户意图。其中，用户意图可能是通过自然语言来表达的，那么则要通过意图解析技术(语义分析技术等)将当前用户意图转化成实际的意图指标，用于计算机识别用户的需求，例如用户意图是要专网实现低时延，通过语义识别技术确定意图指标是传输时延，我们把传输时延称作一个意图指标。可以理解的是，用户意图往往不止一个，这样通过意图解析就可以得到多个意图指标，为了协调各个意图指标对候选优化方案的影响，为意图指标分配各自的权重，各权重之和等于1。在实际应用中，假如存在多个用户意图且用户更偏重于网络带宽，那么则将网络带宽这一意图指标对应的权重调整得更高，以得到贴合用户需求的候选优化方案。It is understandable that no matter what factors cause the mapping relationship not to satisfy the current user intention, the current user intention must be considered when recreating the mapping relationship. If the user intention does not change when generating candidate optimization solutions, Then take it as the current user intention, and if the user intention changes when the candidate optimization scheme is generated, take the changed user intention as the current user intention. Among them, the user's intention may be expressed through natural language, then the current user's intention must be converted into an actual intention indicator through intention analysis technology (semantic analysis technology, etc.), which is used for the computer to identify the user's needs, for example, the user's intention is to The private network achieves low latency, and the intent indicator determined by semantic recognition technology is the transmission delay. We call the transmission delay an intent indicator. It is understandable that there is often more than one user intent, so multiple intent indicators can be obtained through intent analysis. In order to coordinate the impact of each intent indicator on the candidate optimization scheme, the intent indicators are assigned their respective weights, and the sum of each weight is equal to 1. In practical applications, if there are multiple user intentions and the user prefers network bandwidth, then the weight corresponding to the intention index of network bandwidth is adjusted to be higher to obtain a candidate optimization scheme that meets the user's needs.

可以理解的是，基于意图编排(或称意图策略编排)得到映射关系往往十分复杂，因此意图策略编排通常采用人工智能技术进行计算，解决了大量数据编排的问题，同时也提供了较为精确的优化方案。人工智能技术的分支较多，本申请实施例可以采用强化学习的方案，不仅能利用现有数据，还可以通过对网络环境的探索获得新数据，并利用新数据循环往复地更新迭代现有模型，适合于网络模型的决策。为了指示强化学习要达到的目标，设置理想目标函数，该理想目标函数由意图指标和意图指标的权重构建，在意图策略编排过程中通过将意图指标进行量化得到量化指标值，并针对该指标值进行评分，多个意图指标集合可以得到意图指标的总评分；强化学习通过理想目标函数所得到的意图指标的总评分进行评估，从而得到输出得到候选优化方案。It is understandable that the mapping relationship based on intent orchestration (or intent policy orchestration) is often very complicated. Therefore, intent policy orchestration usually uses artificial intelligence technology for calculation, which solves the problem of a large amount of data orchestration and provides more accurate optimization. plan. There are many branches of artificial intelligence technology. The embodiment of this application can adopt the scheme of reinforcement learning, which can not only use the existing data, but also obtain new data through the exploration of the network environment, and use the new data to update and iterate the existing model repeatedly. , suitable for network model decision making. In order to indicate the goal to be achieved by reinforcement learning, an ideal objective function is set. The ideal objective function is constructed by the intention index and the weight of the intention index. For scoring, multiple sets of intent indicators can get the total score of the intent indicators; reinforcement learning evaluates the total score of the intent indicators obtained by the ideal objective function, so as to obtain the output and obtain the candidate optimization scheme.

对于多个意图指标的情况，不同类型的意图指标具有不同的特性，对其进行指标评分的方式也不相同。例如，某条虚拟链路的意图指标为时延大小，相应的量化指标值为时延值，时延值越低，该虚拟链路的性能越好，因此该意图指标的得分越高；又例如，某条链路的意图指标为带宽大小，量化指标值为带宽，带宽越大，该意图指标的得分越高；还有一些意图指标是用于表征整个虚拟网络的服务性能的，例如，全网负载均衡度，需要在全部虚拟链路的意图指标确定得分后才能评估得到得分。因此，本申请实施例将意图指标分为三类，包括第一类意图指标、第二类意图指标和第三类意图指标。For the case of multiple intent indicators, different types of intent indicators have different characteristics, and the ways of scoring them are also different. For example, the intent index of a virtual link is the delay, and the corresponding quantitative index value is the delay value. The lower the delay value, the better the performance of the virtual link, so the higher the score of the intent index; For example, the intent index of a link is the bandwidth, and the quantitative index value is bandwidth. The larger the bandwidth, the higher the score of the intent index; there are also some intent indexes used to characterize the service performance of the entire virtual network, for example, The load balancing degree of the entire network can only be evaluated and obtained after the intent indicators of all virtual links determine the scores. Therefore, the embodiment of the present application divides the intention indicators into three categories, including the first-type intention indicators, the second-type intention indicators, and the third-type intention indicators.

第一类意图指标为虚拟链路经过意图策略编排得到的量化指标值与指标得分成反比的意图指标。The first type of intent index is the intent index whose quantified index value and index score are inversely proportional to the index score obtained by the virtual link through the intent policy orchestration.

第二类意图指标为虚拟链路经过意图策略编排得到的量化指标值与指标得分成正比的意图指标。The second type of intent index is the intention index whose quantitative index value obtained by the virtual link through intent strategy orchestration is proportional to the index score.

第三类意图指标为所有虚拟链路经过意图策略编排后用于表征虚拟网络的业务性能的意图指标。The third type of intent index is the intent index used to represent the service performance of the virtual network after all the virtual links are orchestrated by the intent policy.

虚拟链路经过意图策略编排，得到量化指标值并基于一定的评分基准为该量化指标值打分，通过对虚拟链路的每个意图指标值进行打分，可以汇总得到各条虚拟链路上各个意图指标的得分总和；第三类意图指标可以在对最后一条虚拟链路进行意图策略编排时实时得分评估，打全网服务性能相对应的意图指标打分。将三类意图指标的得分值进行汇总，即可确定理想目标函数的总评分。The virtual link is arranged through the intent strategy to obtain the quantitative index value and score the quantitative index value based on a certain scoring basis. By scoring each intent index value of the virtual link, each intent on each virtual link can be summarized The sum of the scores of the indicators; the third type of intent indicators can be evaluated in real time when the intent policy is programmed for the last virtual link, and scored with the intent indicators corresponding to the service performance of the entire network. The total score of the ideal objective function can be determined by summarizing the score values of the three types of intent indicators.

可以理解的是，为了统一各个意图指标的得分的评分标准，还可以对意图指标值进行归一化处理，归一化方式可以为各个量化指标值预设门限值，根据门限值和意图指标的类型构建归一化算法，从而统一各个意图指标得分。具体的算法举例将在后面示例给出，在此仅简要说明。It is understandable that, in order to unify the scoring standards for the scores of each intent index, the intent index value can also be normalized. The normalization method can preset threshold values for each quantitative index value. According to the threshold value and intent The type of indicator builds a normalization algorithm to unify the scores of each intent indicator. Specific algorithm examples will be given later, and only a brief description is given here.

除了基于用户意图生成候选优化方案，现网的资源情况也是生成候选优化方案的影响因素，其中虚拟网络的业务分布情况和物理网络的拓扑信息分别能够反映当前虚拟网络所提供的业务资源以及当前物理网络中设备和链路的关系。意图策略编排过程中，以上述得到的理想目标函数为优化目标，参考业务分布信息和拓扑信息进行方案构建，自动得到多个候选优化方案。In addition to generating candidate optimization schemes based on user intentions, the resource conditions of the existing network are also influencing factors for generating candidate optimization schemes. The service distribution of the virtual network and the topology information of the physical network can respectively reflect the service resources provided by the current virtual network and the current physical network. The relationship between devices and links in the network. In the process of intent policy orchestration, the ideal objective function obtained above is used as the optimization goal, and the scheme is constructed with reference to the business distribution information and topology information, and multiple candidate optimization schemes are automatically obtained.

可以理解的是，由意图策略编排直接得到候选优化方案并不一定都符合现网要求，例如电信营运商需要保证业务长时间的稳定运行，偏重于稳定性要求，并且需要考虑业务收益和运维支出之间的平衡，因此需要对候选优化方案进行一定的筛选，避免无效的候选优化方案影响后续评估。如步骤S200中的权衡分析方法，参照图3，具体可以通过以下的步骤S210、步骤S220、步骤S230和步骤S240实现。It is understandable that candidate optimization schemes obtained directly from intent policy orchestration do not necessarily meet the requirements of the live network. For example, telecom operators need to ensure long-term stable operation of services, focusing on stability requirements, and need to consider business benefits and operation and maintenance. Therefore, it is necessary to screen candidate optimization schemes to avoid invalid candidate optimization schemes from affecting subsequent evaluations. As for the trade-off analysis method in step S200 , referring to FIG. 3 , it can be specifically implemented through the following steps S210 , S220 , S230 and S240 .

步骤S210，获取候选优化方案的预期增量收益。Step S210, obtaining the expected incremental benefit of the candidate optimization scheme.

步骤S220，获取候选优化方案在执行优化过程带来的预期损失。Step S220, obtaining the expected loss caused by the optimization process of the candidate optimization scheme.

步骤S230，根据预期增量收益和预期损失确定候选优化方案的风险收益比。Step S230, determining the risk-benefit ratio of the candidate optimization scheme according to the expected incremental benefit and expected loss.

步骤S240，当候选优化方案的风险收益比高于风险收益比下限值，并且候选优化方案的预期损失小于损失上限值，确定候选优化方案符合权衡分析门限要求。Step S240, when the risk-benefit ratio of the candidate optimization scheme is higher than the lower limit value of the risk-benefit ratio, and the expected loss of the candidate optimization scheme is less than the upper limit value of the loss, it is determined that the candidate optimization scheme meets the threshold requirement of the trade-off analysis.

预期增量收益表示优化前后两个意图映射策略带来的收益之差，预期损失表示执行候选优化方案造成部分服务终端而带来的损失，预期损失跟以下因素有关。The expected incremental benefit represents the difference between the benefits brought by the two intent mapping strategies before and after optimization, and the expected loss represents the loss caused by some service terminals caused by the execution of the candidate optimization scheme. The expected loss is related to the following factors.

①.因优化而发生重路由、资源重新配置的物理链路的数量。①. The number of physical links where rerouting and resource reconfiguration occur due to optimization.

②.因物理链路重路由而造成的业务终端的损失。②. Loss of service terminals caused by physical link rerouting.

③.因业务中断影响用户体验而带来的损失。③. Losses caused by business interruption affecting user experience.

通过统计候选优化方案带来的预期增量收益和预期损失，可以计算得到候选优化方案的风险收益比，当满足计算得到的风险收益比高于风险收益比下限值，并且满足预期算低于损失上限值，该候选优化方案即符合权衡分析门限要求。通过上述方式解决了网络运营求优和求稳的量化权衡问题，只要其中一个条件不符合，候选优化方案都不能通过。By counting the expected incremental benefits and expected losses brought by candidate optimization schemes, the risk-benefit ratio of the candidate optimization scheme can be calculated. The upper limit value of the loss, the candidate optimization scheme meets the threshold requirements of the trade-off analysis. Through the above method, the quantitative trade-off problem of network operation optimization and stability is solved. As long as one of the conditions is not met, the candidate optimization scheme cannot pass.

将负荷权衡分析门限要求的候选优化方案进行评估，来对比各个符合权衡分析门限要求的候选优化方案之间的优劣，最终从中选取一个作为目标优化方案。例如，参照图4，可以通过以下的步骤S250、步骤S260和步骤S270实现目标优化方案的选取。Evaluate the candidate optimization schemes required by the load balance analysis threshold to compare the pros and cons of each candidate optimization scheme that meets the tradeoff analysis threshold requirements, and finally select one of them as the target optimization scheme. For example, referring to FIG. 4 , the selection of the target optimization scheme can be realized through the following steps S250 , S260 and S270 .

步骤S250，确定符合权衡分析门限要求的候选优化方案的优势衡量条件和优势衡量条件的权重，优势衡量条件包括候选优化方案的权衡分析结果、候选优化方案的优劣程度、当前用户意图中主要意图指标的量化指标值和主要意图指标的门限值中的至少一个。Step S250, determining the advantage measurement conditions and the weights of the advantage measurement conditions of the candidate optimization schemes that meet the requirements of the trade-off analysis threshold. The advantage measurement conditions include the trade-off analysis results of the candidate optimization schemes, the pros and cons of the candidate optimization schemes, and the main intentions of the current user intentions. At least one of the quantitative indicator value of the indicator and the threshold value of the main intent indicator.

步骤S260，根据优势衡量条件和优势衡量条件的权重确定候选优化方案的评估得分。Step S260, determining the evaluation score of the candidate optimization scheme according to the advantage measurement condition and the weight of the advantage measurement condition.

步骤S270，将评估得分最高的候选优化方案作为目标优化方案。Step S270, taking the candidate optimization scheme with the highest evaluation score as the target optimization scheme.

为符合权衡分析门限要求的候选优化方案确定用于评估分数的优势衡量条件，该优势衡量条件表示该候选优化方案的评分项，通常可以选取候选优化方案的权衡分析结果、候选优化方案的优劣程度、当前用户意图中主要意图指标的量化指标值和主要意图指标的门限值中的至少一个；例如，优势衡量条件可以包括收益风险比情况、预期损失情况、理想目标函数得分情况、用户最看重的意图指标、该意图指标的门限值等。并且为了将各个不同类型的优势衡量条件统一到一种标准，对优势衡量条件对应的值进行归一化处理，使优势衡量条件的值统一到0到1之间，另外，还为各个优势衡量条件设置权重值，体现用户意图倾向，基于优势衡量条件和优势衡量条件的权重值，即可计算得到各个候选优化方案的得分。将得分最高的一个候选优化方案作为目标优化方案。Determine the advantage measurement condition for the evaluation score for the candidate optimization scheme that meets the threshold requirements of the trade-off analysis. The advantage measurement condition indicates the scoring item of the candidate optimization scheme. Usually, the trade-off analysis results of the candidate optimization scheme and the pros and cons of the candidate optimization scheme can be selected. at least one of degree, the quantitative index value of the main intention index in the current user intention, and the threshold value of the main intention index; The most important intent indicator, the threshold value of the intent indicator, etc. And in order to unify the different types of advantage measurement conditions into one standard, the values corresponding to the advantage measurement conditions are normalized so that the values of the advantage measurement conditions are unified between 0 and 1. In addition, each advantage measurement The condition sets the weight value to reflect the user's intention tendency. Based on the advantage measurement condition and the weight value of the advantage measurement condition, the score of each candidate optimization scheme can be calculated. The candidate optimization scheme with the highest score is taken as the target optimization scheme.

参照图5，得到目标优化方案后，需要确定执行优化方案的时间段，可以通过以下的步骤S310、步骤S320和步骤S330实现。Referring to FIG. 5 , after obtaining the target optimization plan, it is necessary to determine the time period for executing the optimization plan, which can be realized through the following steps S310 , S320 and S330 .

步骤S310，根据目标优化方案确定优化造成业务中断对应的物理连接数量和优化所需时长。In step S310, the number of physical connections corresponding to service interruption caused by optimization and the time required for optimization are determined according to the target optimization scheme.

步骤S320，根据流量统计函数确定在不同时间窗口内物理连接数量的流量大小，时间窗口的长度不小于优化所需时长。Step S320, according to the traffic statistical function, determine the traffic volume of the physical connection quantity in different time windows, and the length of the time window is not less than the time required for optimization.

步骤S330，选取最小流量对应的时间窗口，作为执行目标优化方案的时间窗口。Step S330, selecting the time window corresponding to the minimum flow rate as the time window for executing the target optimization scheme.

上述步骤描述了选择最佳优化时机的方法，预测得出一个时间窗口T(如果优化与校正是因为预测到未来OTN切片/OVPN专网所映射的物理网络发生故障，那么时间窗口T应该被选择在预测的物理网络的故障发生时间之前)，使得在OTN切片/OVPN专网映射的物理连接中，被因意图优化而造成业务中断的K条物理连接在时间窗口T内的流量总和与其他时间窗口相比最小。这样，这些映射的物理连接因被意图优化而造成流量中断的损失最小。The above steps describe the method of selecting the best optimization opportunity, and predict a time window T (if the optimization and correction is due to the prediction that the physical network mapped to the OTN slice/OVPN private network in the future will fail, then the time window T should be selected Before the predicted failure time of the physical network), so that in the physical connection mapped by the OTN slice/OVPN private network, the total traffic of the K physical connections caused by the intention to optimize the service interruption in the time window T is different from other time The window is the smallest in comparison. In this way, the physical connections of these maps are optimized with minimal penalty for traffic interruption.

此时，时间窗口T内的流量总和可被称为谷底流量，且时间窗口T的长度应大于或等于给定时间段门限长度(执行优化所需的实际时间)，以确保优化操作需要的时间和被优化的物理连接从瞬态的业务中断到进入稳态运行需要的时间。At this time, the sum of the flow in the time window T can be called the bottom flow, and the length of the time window T should be greater than or equal to the threshold length of a given time period (the actual time required to perform optimization) to ensure that the time required for the optimization operation The time required for a physical connection to go from a transient service interruption to steady-state operation is optimized.

通过流量预测等AI技术，可以预测谷底流量值及最近一次时间窗口T的发生时间与持续长度，可在该时间窗口内按照发明内容三的描述对OTN切片/OVPN专网进行意图优化。Through AI technology such as traffic forecasting, the bottom traffic value and the occurrence time and duration of the latest time window T can be predicted, and the OTN slice/OVPN private network can be optimized according to the description of the third content of the invention within this time window.

通过上述各个步骤，可以据用户需求对应的用户意图生成候选优化方案，并基于候选优化方案进行利弊权衡分析，并把握执行候选优化方案的时机，从而实现对虚拟网络的自动优化和校正，使虚拟网络能够自动匹配物理网络资源，满足数字转型时代业务对网络的敏捷性要求。Through the above steps, candidate optimization schemes can be generated according to user intentions corresponding to user needs, and the pros and cons of the candidate optimization schemes can be analyzed based on the trade-offs, and the timing of executing the candidate optimization schemes can be grasped, thereby realizing automatic optimization and correction of the virtual network. The network can automatically match physical network resources to meet the business agility requirements for the network in the era of digital transformation.

下面以实际示例对本申请实施例的虚拟网络的优化方法进行说明。The method for optimizing a virtual network in the embodiment of the present application will be described below with a practical example.

本示例基于OTN切片/OVPN专网，通过四个部分的内容进行说明，分别包括候选优化方案的生成、候选优化方案的权衡分析、候选优化方案的评估和执行优化时机的计算。This example is based on an OTN slice/OVPN private network, and is described in four parts, including generation of candidate optimization solutions, trade-off analysis of candidate optimization solutions, evaluation of candidate optimization solutions, and calculation of execution timing.

(1)候选优化方案的生成(1) Generation of candidate optimization schemes

此处提出了一种表达OTN切片/OVPN专网服务与物理网络资源的意图映射策略的理想目标函数定义方法，假设OTN切片/OVPN专网包含的虚拟链路为m条，按照意图映射策略顺序编排(即在对应的物理网络上顺序建立m条物理连接，与这m条虚链路构成映射关系)，则意图映射策略的理想目标函数可如下定义。Here, an ideal objective function definition method to express the intention mapping strategy of OTN slice/OVPN private network service and physical network resources is proposed. Assuming that OTN slice/OVPN private network contains m virtual links, the order of intent mapping strategy is Orchestration (that is, m physical connections are sequentially established on the corresponding physical network to form a mapping relationship with the m virtual links), the ideal objective function of the intent mapping strategy can be defined as follows.

上式表示意图策略编排所采用的AI算法(如强化学习)在构建候选优化方案过程中用于约束算法的目标函数，通过上述理想目标函数确定意图指标评分总和最高的目标；其中w _i表示第i条经过意图策略编排的虚拟链路的意图指标的综合评分，w _i根据虚拟链路中意图指标的不同类型来确定计算方式，因此下面先介绍当前虚拟网络具有的三种类型的意图指标。 The above formula shows the objective function of the AI algorithm (such as reinforcement learning) adopted by the schematic strategy arrangement to constrain the algorithm in the process of constructing candidate optimization schemes, and the target with the highest sum of intent index scores is determined through the above ideal objective function; where w _i represents the first The comprehensive score of the intent indicators of i virtual links orchestrated by the intent policy, w _i determines the calculation method according to the different types of intent indicators in the virtual link, so the following first introduces the three types of intent indicators in the current virtual network.

设全网业务的意图指标个数为h个，可以分为三类分别以h1、h2和h3表示，分别对应第一类意图指标、第二类意图指标和第三类意图指标，且有h＝h1+h2+h3。Assuming that the number of intent indicators of the whole network business is h, they can be divided into three categories represented by h1, h2 and h3 respectively, corresponding to the first type of intent indicators, the second type of intent indicators and the third type of intent indicators, and h =h1+h2+h3.

h1表示每条OTN虚拟链路经过意图策略编排得到的量化的指标值与指标得分成反比的意图指标的种类个数。h1 indicates the number of types of intent indicators whose quantitative index value and index score are inversely proportional to each OTN virtual link obtained through intent policy orchestration.

h2表示每条OTN虚拟链路经过意图策略编排得到的量化的指标值与指标得分成正比的意图指标的种类个数。h2 represents the number of types of intent indicators whose quantitative index value and index score are proportional to each OTN virtual link obtained through intent policy orchestration.

h3表示只有当完成所有OTN虚拟链路的意图策略编排后才能获得指标得分的意图指标的种类个数(比如全网带宽利用率、误码率、负载均衡度等)，该类意图指标的得分评估，可在对最后一条(即第m条)OTN虚拟链路进行意图策略编排时实施。h3 indicates the number of types of intent indicators (such as network-wide bandwidth utilization, bit error rate, load balance, etc.) that can obtain indicator scores only after the intent policy arrangement of all OTN virtual links is completed. The score of this type of intent indicators The evaluation can be implemented when the intent policy is orchestrated for the last (that is, the mth) OTN virtual link.

假设各个意图指标的权重定义如下，权重值为所有虚拟链路共享。Assume that the weight of each intent indicator is defined as follows, and the weight value is shared by all virtual links.

θ＝(θ ₁,θ ₂,θ ₃,…,θ _h)， θ=(θ ₁ ,θ ₂ ,θ ₃ ,…,θ _h ),

其中，

in,

各个意图指标可如下定义。Each intent indicator may be defined as follows.

P＝(P ₁,P ₂,P ₃,…,P _h)。 P = (P ₁ , P ₂ , P ₃ , . . . , P _h ).

假设各个意图指标的参考门限值定义如下，为所有虚拟链路共享。Assume that the reference thresholds of each intent indicator are defined as follows, which are shared by all virtual links.

P _T＝(P _T1,P _T2,P _T3,…,P _Th)。 P _T = (P _T1 , P _T2 , P _T3 , . . . , P _Th ).

那么，第i条虚拟链路的各个意图指标值如下定义。Then, each intent index value of the i-th virtual link is defined as follows.

P _i＝(P _i1,P _i2,P _i3,…,P _ih1,P _i(h1+1),P _i(h1+2),…,P _i(h1+h2),)， P _i ＝(P _i1 ,P _i2 ,P _i3 ,...,P _ih1 ,P _i(h1+1) ,P _i(h1+2) ,...,P _i(h1+h2), ),

其中，i≠m。Among them, i≠m.

第m条虚拟链路的各个意图指标值如下定义。Each intent index value of the m-th virtual link is defined as follows.

P _m＝(P _m1,P _m2,P _m3,…,P _mh)。 P _m = (P _m1 , P _m2 , P _m3 , . . . , P _mh ).

第i条被意图映射的虚拟链路的意图指标综合评分举例如下。An example of the comprehensive score of the intent index of the i-th virtual link mapped by the intent is as follows.

其中，i≠m。Among them, i≠m.

第m条被意图映射的虚拟链路的意图指标综合评分举例如下。An example of the comprehensive score of the intent index of the m-th virtual link mapped by the intent is as follows.

其中f(P _mn,P _Tn)表示第n个意图指标的得分评估函数，由实际网络情况确定。 Where f(P _mn ,P _Tn ) represents the score evaluation function of the nth intent indicator, which is determined by the actual network situation.

通过上述各意图指标值与自身意图指标参考门限比值的形式，实现意图指标评分的归一化处理(归一化处理可以有多种其它方法，本专利以上述内容为例)。用户对OTN切片/OVPN专网服务的运维意图要求，可通过意图映射策略目标函数的优化权重值体现到对应的策略编排过程中。例如，用户希望当前专网的综合时延尽量低，AI技术可事先将表示时延对应的意图指标的优化权重调高、其他意图指标的优化权重调低，进而在实现意图策略编排的AI算法中将时延指标的评估得分在整个专网意图评分总和中占比增大The normalization processing of the intent index score is realized through the form of the ratio of each intention index value above and the reference threshold value of the own intention index (the normalization processing can have many other methods, and this patent takes the above content as an example). The user's operation and maintenance intent requirements for OTN slice/OVPN private network services can be reflected in the corresponding policy orchestration process through the optimized weight value of the intent mapping policy objective function. For example, if users want the comprehensive delay of the current private network to be as low as possible, AI technology can increase the optimization weight of the intent index corresponding to the delay and lower the optimization weight of other intent indicators in advance, and then realize the AI algorithm of intent policy arrangement The evaluation score of the general delay index accounts for an increasing proportion of the total score of the entire private network intent

(2)候选优化方案的权衡分析(2) Trade-off analysis of candidate optimization schemes

根据上述第一部分内容的描述和定义，采用AI技术得到多个候选优化方案，那么根据网络运营需求先对各个候选优化方案进行初步筛选。因此权衡分析方式如下。According to the description and definition of the first part above, AI technology is used to obtain multiple candidate optimization schemes, and each candidate optimization scheme is initially screened according to network operation requirements. The trade-off analysis is therefore done as follows.

OTN切片/OVPN专网意图优化与校正的权衡判据必须满足下式。The trade-off criterion for OTN slice/OVPN private network intent optimization and correction must satisfy the following formula.

且有Rsk _loss<Rsk _upthd。 And there is Rsk _loss < Rsk _upthd .

其中OBRR表示某一时段T内、采用意图映射策略进行OTN切片/OVPN专网意图映射优化的收益风险比，Opt _bft表示和当前的实际意图映射关系相比，OTN切片/OVPN专网意图映射优化能带来的增量收益，可以采用网络指标值(例如全网时延下降值的多少)等方式表示，Rsk _loss表示因OTN切片/OVPN专网意图映射优化造成OTN切片/OVPN专网部分服务中断而带来的损失，OBRR _downhd表示OTN切片/OVPN专网意图映射优化的收益风险比下限(判断是否值得优化，若OBRR低于该值，则不值得优化)，Rsk _upthd表示OTN切片/OVPN专网意图映射优化的风险损失上限(超过上限，再多的收益也不做优化处理)。 Among them, OBRR represents the profit-risk ratio of OTN slicing/OVPN private network intent mapping optimization using the intent mapping strategy within a certain period of time T, and Opt _bft represents the OTN slicing/OVPN private network intent mapping optimization compared with the current actual intent mapping relationship. The incremental benefits that can be brought can be expressed in the form of network index values (such as the value of the entire network delay drop), and Rsk _loss indicates that some services in the OTN slice/OVPN private network are caused by the OTN slice/OVPN private network intent mapping optimization. Loss caused by interruption, OBRR _downhd indicates the lower limit of the profit-risk ratio of OTN slice/OVPN private network intent mapping optimization (judging whether it is worth optimizing, if OBRR is lower than this value, it is not worth optimizing), Rsk _upthd indicates OTN slice/OVPN The upper limit of the risk loss for private network intent mapping optimization (if the upper limit is exceeded, no amount of profit will be optimized).

通过上述提出的OTN切片/OVPN专网意图优化与校正的利弊权衡分析方法，就是将优化带来的收益与风险比作为优化操作与否的判据，解决了网络运营求优与求稳的量化权衡问题，即：当OTN切片/OVPN专网意图映射优化的实施方法满足上述的权衡判据要求时，就进行OTN切片/OVPN专网意图映射优化实施处理，否则不做处理。Through the pros and cons trade-off analysis method of OTN slice/OVPN private network intent optimization and correction proposed above, the profit and risk ratio brought by optimization is used as the criterion for optimization operation or not, and the quantification of network operation optimization and stability is solved. The trade-off problem, that is: when the implementation method of OTN slice/OVPN private network intent mapping optimization meets the above-mentioned trade-off criterion requirements, the OTN slice/OVPN private network intent mapping optimization implementation process will be performed, otherwise no processing will be performed.

(3)候选优化方案的评估(3) Evaluation of candidate optimization schemes

通过第一部分内容和第二部分内容得到符合权衡分析门限要求的候选优化方案，本部分内容对符合权衡分析门限要求的候选优化方案的评估得分进行计算，从而选出最优优化方案。为了方便对评估过程进行说明，下面以实际例子列举两个候选优化方案，并直接计算两个候选优化方案的评估得分。Through the content of the first part and the second part, the candidate optimization scheme that meets the threshold requirements of the trade-off analysis is obtained. This part calculates the evaluation scores of the candidate optimization schemes that meet the threshold requirements of the trade-off analysis, so as to select the optimal optimization scheme. In order to facilitate the description of the evaluation process, two candidate optimization schemes are listed below with practical examples, and the evaluation scores of the two candidate optimization schemes are directly calculated.

参照图6，网络切片A中的6条虚拟链路与物理网络的映射关系如图6所示，以VL1-VL6表示6条虚拟链路，Path1-Path6分别表示6条物理连接，VL1-VL6和Path1-Path6一一对应。按照第一部分内容，可以确定生成候选优化方案所采用的理想目标函数如所示。Referring to Figure 6, the mapping relationship between the six virtual links and the physical network in network slice A is shown in Figure 6. VL1-VL6 represent the six virtual links, Path1-Path6 represent the six physical connections, and VL1-VL6 One-to-one correspondence with Path1-Path6. According to the content of the first part, the ideal objective function used to generate candidate optimization schemes can be determined as shown.

其中，设定时延的意图指标的权重为0.6，该意图指标被作为主要意图指标加以优化考虑，即θ _latency＝0.6；假设要求6条虚拟链路映射路径的时延门限为15ms，即要求6条虚拟链路映射路径的时延总和必须小于等于15ms，可以如下表示。 Among them, the weight of the intent index of the delay is set to 0.6, and this intent index is considered as the main intent index for optimization, that is, θ _latency = 0.6; assuming that the delay threshold of the six virtual link mapping paths is required to be 15ms, that is, the required The total delay of the six virtual link mapping paths must be less than or equal to 15ms, which can be expressed as follows.

6*P _i-latency≤P _T-latency＝15ms， 6*P _i- latency ≤ P _T-latency = 15ms,

其中P _i-latency表示第i条虚拟链路的时延指标值，P _T-latency表示切片A的时延门限。 Among them, P _i-latency represents the delay index value of the i-th virtual link, and PT _-latency represents the delay threshold of slice A.

基于故障预测、意图校验等原因需要对VL4虚链路映射的Path4的AB段链路、VL6虚链路映射的Path6的SN段链路进行重优化，参照图7中标记为“×”的虚拟路径，基于上述理想目标函数，采用AI算法获得两种OTN切片/OVPN专网优化方案，分别记为方案一和方案二，如图8和图9所示，以点线表示重优化方案对应的物理连接。Based on reasons such as fault prediction and intention verification, it is necessary to re-optimize the AB segment link of Path4 mapped to the VL4 virtual link, and the SN segment link of Path6 mapped to the VL6 virtual link. Refer to the "×" in Figure 7 Virtual path, based on the above ideal objective function, adopts AI algorithm to obtain two OTN slice/OVPN private network optimization schemes, which are respectively recorded as scheme 1 and scheme 2, as shown in Figure 8 and Figure 9, and the corresponding re-optimization scheme is represented by a dotted line physical connection.

依据第二部分内容对方案一和方案二进行权衡分析，并假设方案一和方案二均通过了权衡分析。在权衡分析过程中，可以得到收益风险比、预期损失等参数。According to the content of the second part, the trade-off analysis is carried out on the scheme 1 and the scheme 2, and it is assumed that the scheme 1 and the scheme 2 both pass the trade-off analysis. In the process of trade-off analysis, parameters such as profit-risk ratio and expected loss can be obtained.

设定本部分内容中优势衡量条件包括所述候选优化方案的权衡分析结果、候选优化方案的优劣程度、当前用户意图中主要意图指标的量化指标值和所述主要意图指标的门限值(具体参数可以参照下面表1)；对上述优势衡量条件，采用如下公式对两个方案进行优势评估得分计算。Setting the advantage measurement conditions in this part includes the trade-off analysis results of the candidate optimization scheme, the pros and cons of the candidate optimization scheme, the quantitative index value of the main intention index in the current user intention, and the threshold value of the main intention index ( For the specific parameters, please refer to the following table 1); for the above-mentioned advantage measurement conditions, use the following formula to calculate the advantage evaluation score of the two schemes.

ρ _i＝Coff(δ _i)， ρ _i =Coff(δ _i ),

其中0<δ _i≤1，δ _i表示作为候选优化方案的优势衡量条件i的真实值，如时延为30ms，或目标函数得分为50分等，ρ _i表示经过Coff函数归一化处理后的优势衡量条件i对应的评估系数，是取值为(0,1]区间的实数。 Where 0<δ _i ≤ 1, δ _i represents the true value of the advantage measurement condition i as a candidate optimization scheme, such as a delay of 30 ms, or an objective function score of 50, etc., ρ _i represents the normalized value of the Coff function The evaluation coefficient corresponding to the superiority measurement condition i is a real number whose value is in the interval (0,1].

基于上述设定，利用下式计算优势评估得分。Based on the above settings, the dominance evaluation score was calculated using the following formula.

F _elv＝∑σ _i·ρ _i， F _elv =∑σ _i ·ρ _i ,

其中∑σ _i＝1，σ _i表示优势衡量条件i的权重，可根据用户对方案衡量条件偏好，调整对应衡量条件i的优势评估权重值；F _elv表示当前候选优化方案的评估得分，实际计算结果是取值为(0,1]区间的实数。 Where ∑σ _i = 1, σ _i represents the weight of the advantage measurement condition i, and the weight value of the advantage evaluation corresponding to the measurement condition i can be adjusted according to the user’s preference for the measurement condition i; F _elv represents the evaluation score of the current candidate optimization plan, the actual The result is a real number whose value is in the interval (0,1].

采用上述公式，在两个候选优化方案均满足“大于收益风险比下限”、“不超过风险损失上限”等严格约束条件前提下，得出两和候选优化方案的评估对比如表1所示，最终选择方案一作为最优优化方案。Using the above formula, under the premise that the two candidate optimization schemes satisfy strict constraints such as "greater than the lower limit of the benefit-risk ratio" and "not exceeding the upper limit of the risk loss", the evaluation comparison of the two candidate optimization schemes is shown in Table 1. Finally, option one is selected as the optimal optimization solution.

表1.意图映射优化的两种方案的评估对比Table 1. Evaluation comparison of two schemes for intent mapping optimization

(4)执行优化时机的计算(4) Execute the calculation of optimization timing

本部分内容描述了一种决策意图优化最佳时机的方法，预测得出一个时间窗口T(如果优化与校正是因为预测到未来OTN切片/OVPN专网所映射的物理网络发生故障，那么时间窗口T应该被选择在预测的物理网络的故障发生时间之前)，使得在OTN切片/OVPN专网映射的物理连接中，被因意图优化而造成业务中断的K条物理连接在时间窗口T内的流量总和与其他时间窗口相比最小。这样，这些映射的物理连接因被意图优化而造成流量中断的损失最小。This part describes a method of decision-making intent to optimize the best time, and predicts a time window T (if the optimization and correction is due to the prediction that the physical network mapped to the OTN slice/OVPN private network in the future will fail, then the time window T should be selected before the predicted failure time of the physical network), so that in the physical connection mapped by the OTN slice/OVPN private network, the traffic of the K physical connections that are interrupted due to the intention to optimize the service within the time window T The sum is minimal compared to other time windows. In this way, the physical connections of these maps are optimized with minimal penalty for traffic interruption.

时间窗口T内的流量总和可被称为谷底流量，且时间窗口T的长度应大于或等于给定时间段门限长度(与执行优化所需的实际时间有关)，以确保优化操作需要的时间和被优化的物理连接从瞬态的业务中断到进入稳态运行需要的时间。参照图10所示，因此时间窗口T的时机把握可用公式描述如下。The sum of the traffic in the time window T can be called the bottom flow, and the length of the time window T should be greater than or equal to the threshold length of a given time period (related to the actual time required to perform optimization), so as to ensure that the time required for the optimization operation and The time required for an optimized physical connection to transition from a transient service interruption to steady-state operation. Referring to FIG. 10 , the timing of the time window T can be described by the formula as follows.

出现谷底流量的时间窗口T满足下式。The time window T in which valley flow occurs satisfies the following formula.

其中T>T _threshold，T _threshold表示设定的时间段门限长度，T＝t2-t1，那么可得下式。 Where T>T _threshold , T _threshold represents the set time threshold length, T=t2-t1, then the following formula can be obtained.

通过AI技术，预测谷底流量值及最近一次时间窗口T的发生时间与持续长度，可在该时间窗口内按照第三部分内容的描述对OTN切片/OVPN专网进行意图优化。Through AI technology, the bottom traffic value and the occurrence time and duration of the latest time window T can be predicted, and the OTN slice/OVPN private network can be optimized according to the description in the third part within this time window.

本申请实施例还提供了一种虚拟网络的优化装置，包括至少一个处理器和用于与至少一个处理器通信连接的存储器；存储器存储有能够被至少一个处理器执行的指令，指令被至少一个处理器执行，以使至少一个处理器能够执行前述的虚拟网络的优化方法。The embodiment of the present application also provides a device for optimizing a virtual network, including at least one processor and a memory for communicating with the at least one processor; the memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor Execution by a processor, so that at least one processor can execute the aforementioned virtual network optimization method.

参照图11，以虚拟网络的优化装置1000中的控制处理器1001和存储器1002可以通过总线连接为例。存储器1002作为一种非暂态计算机可读存储介质，可用于存储非暂态软件程序以及非暂态性计算机可执行程序。此外，存储器1002可以包括高速随机存取存储器，还可以包括非暂态存储器，例如至少一个磁盘存储器、闪存器件、或其他非暂态固态存储器件。在一些实施方式中，存储器1002可包括相对于控制处理器1001远程设置的存储器，这些远程存储器可以通过网络连接至虚拟网络的优化装置1000。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。Referring to FIG. 11 , it is taken as an example that the control processor 1001 and the memory 1002 in the virtual network optimization apparatus 1000 can be connected through a bus. The memory 1002, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory 1002 may include a high-speed random access memory, and may also include a non-transitory memory, such as at least one disk memory, a flash memory device, or other non-transitory solid-state storage devices. In some implementations, the storage 1002 may include storages that are located remotely relative to the control processor 1001 , and these remote storages may be connected to the optimization device 1000 of the virtual network through a network. Examples of the aforementioned networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

本领域技术人员可以理解，图11中示出的装置结构并不构成对虚拟网络的优化装置1000的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置。Those skilled in the art can understand that the device structure shown in FIG. 11 does not constitute a limitation to the virtual network optimization device 1000, and may include more or less components than those shown in the figure, or combine certain components, or have different Part placement.

本申请实施例还提供了一种计算机可读存储介质，该计算机可读存储介质存储有计算机可执行指令，该计算机可执行指令被一个或多个控制处理器执行，例如，被图11中的一个控制处理器1001执行，可使得上述一个或多个控制处理器执行上述方法实施例中的虚拟网络的优化方法，例如，执行以上描述的图1中的方法步骤S100至步骤S400、图2中的方法步骤S110至步骤S140、图3中的方法步骤S210至步骤S240、图4中的方法步骤S250至步骤S270和图5中的方法步骤S310至步骤S330。The embodiment of the present application also provides a computer-readable storage medium, the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are executed by one or more control processors, for example, by the Execution by one control processor 1001 can cause the above-mentioned one or more control processors to execute the optimization method of the virtual network in the above-mentioned method embodiment, for example, execute the method steps S100 to S400 in FIG. 1 described above, and the method in FIG. 2 The method step S110 to step S140, the method step S210 to step S240 in FIG. 3 , the method step S250 to step S270 in FIG. 4 and the method step S310 to step S330 in FIG. 5 .

本申请实施例提供的虚拟网络的优化方法，至少具有如下有益效果：本申请实施例提供了基于用户意图优化和校正虚拟网络与物理网络资源的映射关系的解决方案，根据用户需求对应的用户意图生成候选优化方案，并基于候选优化方案进行利弊权衡分析，并把握执行候选优化方案的时机，从而实现对虚拟网络的自动优化和校正，使虚拟网络能够自动匹配物理网络资源，满足数字转型时代业务对网络的敏捷性要求。The virtual network optimization method provided by the embodiment of the present application has at least the following beneficial effects: The embodiment of the present application provides a solution for optimizing and correcting the mapping relationship between the virtual network and the physical network resources based on user intentions, and according to user intentions corresponding to user needs Generate candidate optimization schemes, conduct pros and cons trade-off analysis based on candidate optimization schemes, and grasp the timing of implementing candidate optimization schemes, so as to realize automatic optimization and correction of virtual networks, so that virtual networks can automatically match physical network resources to meet business requirements in the digital transformation era Agility requirements for the network.

以上所描述的装置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，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。The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

本领域普通技术人员可以理解，上文中所公开方法中的全部或某些步骤、***可以被实施为软件、固件、硬件及其适当的组合。某些物理组件或所有物理组件可以被实施为由处理器，如中央处理器、数字信号处理器或微处理器执行的软件，或者被实施为硬件，或者被实施为集成电路，如专用集成电路。这样的软件可以分布在计算机可读介质上，计算机可读介质可以包括计算机存储介质(或非暂时性介质)和通信介质(或暂时性介质)。如本领域普通技术人员公知的，术语计算机存储介质包括在用于存储信息(诸如计算机可读指令、数据结构、程序模块或其他数据)的任何方法或技术中实施的易失性和非易失性、可移除和不可移除介质。计算机存储介质包括但不限于RAM、ROM、EEPROM、闪存或其他存储器技术、CD-ROM、数字多功能盘(DVD)或其他光盘存储、磁盒、磁带、磁盘存储或其他磁存储装置、或者可以用于存储期望的信息并且可以被计算机访问的任何其他的介质。此外，本领域普通技术人员公知的是，通信介质通常包含计算机可读指令、数据结构、程序模块或者诸如载波或其他传输机制之类的调制数据信号中的其他数据，并且可包括任何信息递送介质。Those skilled in the art can understand that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware and an appropriate combination thereof. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit . Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). As known to those of ordinary skill in the art, the term computer storage media includes both volatile and nonvolatile media implemented in any method or technology for storage of information, such as computer readable instructions, data structures, program modules, or other data. permanent, removable and non-removable media. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cartridges, tape, magnetic disk storage or other magnetic storage devices, or can Any other medium used to store desired information and which can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media .

以上是对本申请的若干实施方式进行了具体说明，但本申请并不局限于上述实施方式，熟悉本领域的技术人员在不违背本申请精神的前提下还可作出种种的等同变形或替换，这些等同的变形或替换均包含在本申请权利要求所限定的范围内。The above is a specific description of several embodiments of the present application, but the present application is not limited to the above-mentioned embodiments, and those skilled in the art can also make various equivalent deformations or replacements without violating the spirit of the present application. Equivalent modifications or replacements are all within the scope defined by the claims of the present application.

Claims

一种虚拟网络的优化方法，所述虚拟网络的虚拟链路与物理网络的物理网络资源构成映射关系，其中，所述优化方法包括：A method for optimizing a virtual network, wherein a virtual link of the virtual network forms a mapping relationship with a physical network resource of a physical network, wherein the optimization method includes:

在所述映射关系不满足当前用户意图的情况下，基于当前用户意图生成至少两个候选优化方案；In the case that the mapping relationship does not satisfy the current user intention, at least two candidate optimization schemes are generated based on the current user intention;

对所述候选优化方案进行权衡分析，并从符合权衡分析门限要求的候选优化方案中确定目标优化方案，所述权衡分析为通过门限对所述候选优化方案进行筛选的过程；Performing a tradeoff analysis on the candidate optimization schemes, and determining a target optimization scheme from the candidate optimization schemes that meet the threshold requirements of the tradeoff analysis, the tradeoff analysis is a process of screening the candidate optimization schemes through a threshold;

根据所述物理网络的流量变化规律确定执行所述目标优化方案的优化时间窗口；determining an optimization time window for executing the target optimization scheme according to the traffic variation law of the physical network;

根据所述目标优化方案和所述优化时间窗口重新建立映射关系。A mapping relationship is re-established according to the target optimization solution and the optimization time window.
根据权利要求1所述的虚拟网络的优化方法，其中，所述映射关系不满足当前用户意图，包括以下至少之一的情况：The method for optimizing a virtual network according to claim 1, wherein the mapping relationship does not satisfy the current user intention, including at least one of the following situations:

由用户需求的变化而引起用户意图的变化；Changes in user intent caused by changes in user needs;

或所述虚拟链路发生调整而引起当前的映射关系无效；Or the adjustment of the virtual link causes the current mapping relationship to be invalid;

或所述物理网络资源发生调整或性能劣化而引起当前的映射关系无效；Or the adjustment or performance degradation of the physical network resources causes the current mapping relationship to be invalid;

或预测所述物理网络发生故障而导致当前的映射关系即将无效。Or it is predicted that the physical network fails and the current mapping relationship is about to become invalid.
根据权利要求1所述的虚拟网络的优化方法，其中，所述基于当前用户意图生成至少两个候选优化方案，包括：The method for optimizing a virtual network according to claim 1, wherein said generating at least two candidate optimization schemes based on current user intentions comprises:

获取当前用户意图、所述虚拟网络的业务分布信息和所述物理网络的拓扑信息；Acquiring current user intentions, service distribution information of the virtual network, and topology information of the physical network;

根据所述用户意图确定意图指标和所述意图指标的权重；determining an intent indicator and a weight of the intent indicator according to the user intent;

根据所述意图指标和所述意图指标的权重构建理想目标函数；constructing an ideal objective function according to the weight of the intention index and the intention index;

基于所述理想目标函数和所述业务分布信息、所述拓扑信息，通过意图策略编排得到至少两个候选优化方案。Based on the ideal objective function, the service distribution information, and the topology information, at least two candidate optimization schemes are obtained through intent policy orchestration.
根据权利要求3所述的虚拟网络的优化方法，其中，所述意图指标包括第一类意图指标、第二类意图指标和第三类意图指标；The method for optimizing a virtual network according to claim 3, wherein the intention indicators include a first type of intention indicator, a second type of intention indicator, and a third type of intention indicator;

所述第一类意图指标为虚拟链路经过意图策略编排得到的量化指标值与指标得分成反比的意图指标；The first type of intent index is an intent index whose quantitative index value and index score are inversely proportional to the index score obtained by the virtual link through intent policy arrangement;

所述第二类意图指标为虚拟链路经过意图策略编排得到的量化指标值与指标得分成正比的意图指标；The second type of intent index is an intent index whose quantitative index value and index score are directly proportional to the index score obtained by the virtual link through intent strategy arrangement;

所述第三类意图指标为所有虚拟链路经过意图策略编排后用于表征所述虚拟网络的业务性能的意图指标。The third type of intent index is an intent index used to characterize the service performance of the virtual network after all virtual links are orchestrated by the intent policy.
根据权利要求1所述的虚拟网络的优化方法，其中，所述对所述候选优化方案进行权衡分析，包括：The method for optimizing a virtual network according to claim 1, wherein said performing a trade-off analysis on said candidate optimization scheme includes:

获取所述候选优化方案的预期增量收益；Obtaining the expected incremental benefit of the candidate optimization scheme;

获取所述候选优化方案在执行优化过程带来的预期损失；Obtaining the expected loss caused by the candidate optimization scheme during the execution of the optimization process;

根据所述预期增量收益和所述预期损失确定所述候选优化方案的风险收益比；determining the risk-benefit ratio of the candidate optimization scheme according to the expected incremental benefit and the expected loss;

当所述候选优化方案的风险收益比高于风险收益比下限值，并且所述候选优化方案的预期损失小于损失上限值，确定所述候选优化方案符合权衡分析门限要求。When the risk-benefit ratio of the candidate optimization scheme is higher than the lower limit value of the risk-benefit ratio, and the expected loss of the candidate optimization scheme is less than the loss upper limit value, it is determined that the candidate optimization scheme meets the trade-off analysis threshold requirement.
根据权利要求1所述的虚拟网络的优化方法，其中，所述从符合权衡分析门限要求的候选优化方案中确定目标优化方案，包括：The method for optimizing a virtual network according to claim 1, wherein said determining the target optimization scheme from the candidate optimization schemes that meet the trade-off analysis threshold requirements includes:

确定符合权衡分析门限要求的候选优化方案的优势衡量条件和所述优势衡量条件的权重，所述优势衡量条件包括所述候选优化方案的权衡分析结果、所述候选优化方案的优劣程度、当前用户意图中主要意图指标的量化指标值和所述主要意图指标的门限值中的至少一个；Determine the advantage measurement conditions and the weights of the advantage measurement conditions of the candidate optimization schemes that meet the tradeoff analysis threshold requirements, the advantage measurement conditions include the tradeoff analysis results of the candidate optimization schemes, the pros and cons of the candidate optimization schemes, the current At least one of the quantitative index value of the main intention indicator in the user intention and the threshold value of the main intention indicator;

根据所述优势衡量条件和所述优势衡量条件的权重确定所述候选优化方案的评估得分；determining the evaluation score of the candidate optimization scheme according to the advantage measuring condition and the weight of the advantage measuring condition;

将评估得分最高的候选优化方案作为目标优化方案。The candidate optimization solution with the highest evaluation score is taken as the target optimization solution.
根据权利要求1所述的虚拟网络的优化方法，其中，所述根据所述物理网络的流量变化规律确定执行所述目标优化方案的优化时间窗口，包括：The method for optimizing a virtual network according to claim 1, wherein said determining an optimization time window for executing the target optimization scheme according to the flow change law of the physical network comprises:

根据所述目标优化方案确定优化造成业务中断对应的物理连接数量和优化所需时长；Determine the number of physical connections corresponding to business interruption caused by optimization and the time required for optimization according to the target optimization scheme;

根据流量统计函数确定在不同时间窗口内所述物理连接数量的流量大小，所述时间窗口的长度不小于所述优化所需时长；Determine the traffic size of the number of physical connections in different time windows according to the traffic statistics function, and the length of the time window is not less than the time required for the optimization;

选取最小流量对应的时间窗口，作为执行所述目标优化方案的时间窗口。The time window corresponding to the minimum flow rate is selected as the time window for executing the target optimization scheme.
根据权利要求7所述的虚拟网络的优化方法，其中，在预测所述物理网络发生故障而导致当前的映射关系即将无效的情况下，执行所述目标优化方案的时间窗口早于预测发生故障的时间点。The method for optimizing a virtual network according to claim 7, wherein, in the case that the current mapping relationship is about to become invalid due to a failure of the physical network predicted, the time window for executing the target optimization solution is earlier than the predicted failure point in time.
一种虚拟网络的优化装置，包括至少一个处理器和用于与所述至少一个处理器通信连接的存储器；其中，所述存储器存储有能够被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如权利要求1至8中任意一项所述的虚拟网络的优化方法。An optimization device for a virtual network, comprising at least one processor and a memory for communicating with the at least one processor; wherein, the memory stores instructions that can be executed by the at least one processor, and the instructions Executed by the at least one processor, so that the at least one processor can execute the virtual network optimization method according to any one of claims 1-8.
一种计算机可读存储介质，其中，所述计算机可读存储介质存储有计算机可执行指令，所述计算机可执行指令用于使计算机执行如权利要求1至8中任意一项所述的虚拟网络的优化方法。A computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the virtual network according to any one of claims 1 to 8 optimization method.