CN116074389A - Wide area super fusion service customization system facing distributed cloud - Google Patents

Abstract

The invention provides a wide area super fusion service customizing system facing a distributed cloud, which comprises the following steps: constructing a distributed cloud-oriented wide area super fusion service customization network architecture based on forwarding, control and business three planes; a novel protocol stack layered structure centering on service is created, network capacity of a service customization network is rearranged, and the novel protocol stack layered structure comprises an application layer, a service expression layer, a service transmission layer, a service interconnection layer, a packet bearing layer, a data link layer and a physical layer; meanwhile, the supply and demand matching mechanism of the application side and the network side is constructed by referring to the end-to-end travel service concept of the travel traffic third party platform. The service customization network is introduced into the distributed cloud, so that the service networking capability of the cross-wide area network and through distributed cloud, deterministic transmission, differentiated requirements of each application, network service capability used according to the needs and efficient coordination between the network and the application are realized.

Description

Wide area super fusion service customization system facing distributed cloud

Technical Field

The invention relates to the technical field of communication, in particular to a distributed cloud-oriented wide area super fusion service customization system.

Background

Along with the fusion of information technology (Information Technology, IT) and computer technology (Computer Technology, CT), cloud network fusion becomes a consistent consensus and important development trend, and the core idea is to fuse computing and storage functions in a network and realize the collaborative management and scheduling of multidimensional resources such as network computing and storage by using a software-defined mode. Particularly, with the development of technologies such as edge computing, power networks (Computing Power Network, CPN) and the like in recent years, the concept of "distributed cloud" is proposed and enhanced, and the cloud network integration is further enhanced by Gartner listed in ten strategic technological development trends in 2019 and 2020. The core idea of distributed cloud is to scatter and sink the service capabilities of cloud computing from a high-level single point in the network to a low-level multiple point in the network to form a wide-area and large-scale distributed cloud service capability. However, in a distributed cloud, the connection at the cloud edge needs to span a wide area network. The traditional transmission control protocol/internet protocol (Transmission Control Protocol/Internet Protocol, TCP/IP) technical system has the problems that the transmission quality is difficult to expect due to the statistical multiplexing of forwarding resources, the network is manageable and uncontrollable due to the distributed routing control, the application network is weak and cooperative due to the end-to-end application transmission, and the like, so that the cloud side ends can be connected together through the unified control of the application, but the efficient data flow is difficult to realize through the network, and the cloud side ends cannot really realize the organic fusion.

Service customization networks (Service Customized Network, SCN) are a new model of future network architecture autonomously proposed by Liu Yunjie institution team. The concept of SCN was proposed in 2014, and in combination with software-defined networks, network function virtualization, and cloud computing technologies, an attempt was made to construct a network architecture with differentiated service capabilities that is open, scalable, efficient, and flexible in scheduling information resources, so that a network infrastructure provider can implement flexible scheduling of network resources such as computing, storage, and bandwidth according to the personalized needs of users, and construct a virtual network with different service capabilities, and further construct a new industrial ecological chain.

The service customization network is introduced into the distributed cloud, so that the traditional modeling mode of taking the resources as the center can be changed, and the global resources are subjected to multi-target joint scheduling through the thought of taking the services as the center, so that the super-fusion integration of network, calculation and storage is realized. However, introducing service customization networks in distributed clouds faces a number of technical challenges, including:

(1) Service semantics are not reflected in the whole layering of TCP/IP, and the service quality capability provided in a 'patching' mode can not fundamentally change the problems of 'transmission quality is difficult to expect' and 'network manageability is uncontrollable'.

(2) The network application collaboration is a key problem for realizing the super fusion network, and the traditional network is oriented to a host, identifies the application through deep packet inspection and imports a corresponding network path, but is difficult to be adopted by an application party because the network party needs to invade the interior of the application; network path control can also be directly accomplished by application-initiated source routing, however, the network party needs to open control rights to the application party and is difficult to accept by the network party. The network and the application are relatively independent, the network side and the application side belong to different main bodies, and a huge gap exists between the network and the application.

Disclosure of Invention

In view of this, the embodiment of the invention provides a wide area super fusion service customization system facing to a distributed cloud, so as to eliminate or improve one or more defects existing in the prior art, and solve the problems that transmission quality is difficult to expect, network management is uncontrollable and efficient cooperation between a network and an application is not possible when a service customization network is fused in the distributed cloud in the prior art.

The invention provides a distributed cloud-oriented wide-area super-fusion service customization system, which is characterized by comprising the following components:

the application layer is used for carrying out data encapsulation, encryption and decryption on the data packets of each application;

The service expression layer is used for carrying out multidimensional service identification addressing on each application and the attribute thereof of the application layer, wherein the attribute refers to the characteristic attached to each application in communication;

the service transmission layer adopts an open interface preset by the white box switch to control the transmission of the service customized network access side;

the service interconnection layer adopts a software defined network controller to realize the connection communication between the service customization network and each wide area network;

the packet bearing layer adopts a segmented router to forward according to a network path of a designated data packet at each node of the system; the packet bearing layer encapsulates the forwarding capability into NaaS form, matches the forwarding capability with the requirements of each application, and constructs a consumption matching mechanism;

the data link layer is used for adding a frame head and a frame tail at the head and the tail of the data packet respectively, sealing the frame, and carrying out end-to-end transmission on the data packet by adopting a time sensitive network switch;

the physical layer adopts an Ethernet photoelectric interface for communication; the physical layer provides a multidimensional resource pool constructed by a core cloud server and an edge cloud server, wherein the multidimensional resource pool at least comprises network resources, computing resources, storage resources and time resources.

In some embodiments of the present invention, the service expression layer addresses the multidimensional service identifier for each application and its attribute of the application layer, and further includes:

on the basis of the multi-dimensional service identification, multi-dimensional resources in the multi-dimensional resource pool are subjected to hierarchical scheduling, a target resource cluster is obtained through screening according to a first-level scheduling index, a target node is obtained through screening according to a second-level scheduling index, and tasks of corresponding applications are distributed to the target node; the first-level scheduling index comprises geographic position, use cost and cloud server information; the second-level scheduling index comprises a resource duty ratio, power consumption and dependent association parameters between the resource and the corresponding node.

In some embodiments of the present invention, the communication requirements of each application are analyzed at the application layer, a requirement correlation matrix is generated, a representation paradigm of the communication requirements is constructed, and the communication requirements are expressed through the service expression layer.

In some embodiments of the present invention, relevant factors affecting resource scheduling are abstracted into multidimensional vectors, orthogonal decomposition is performed on the multidimensional vectors to obtain decision subsets of different dimensions, a multi-level decision tree is constructed, and scheduling decisions are generated by using the multi-level decision tree; the related factors include resource duty cycle, power consumption, dependent correlation parameters between the resource and the corresponding node, and resource energy consumption.

In some embodiments of the present invention, a real-time queuing model is constructed based on time slot constraint conditions and with jitter and time delay of a network path as targets, and a queuing scheme corresponding to an application data packet is generated by using the real-time queuing model; the time slot constraint conditions comprise data packet size, number, packet sending time and time resource.

In some embodiments of the present invention, a method of single-path telescoping and multi-path sharing is adopted to elastically shift the network path bandwidth so as to receive burst data.

In some embodiments of the present invention, network state information obtained by analysis feedback is collected in real time, a network optimization model is constructed, and the service customization network is optimized; the network state information includes link load, link bandwidth, link delay, link jitter, link packet loss rate, and link utilization.

In some embodiments of the present invention, a method of mixing probe dial testing and in-band telemetry is adopted to observe the actual delay and jitter of a network path, and the method of injecting and extracting metadata into a production environment message is adopted to sense the network state information in real time.

In some embodiments of the present invention, a network actual consumption model is constructed according to the network state information obtained by real-time sensing, a consumption record of a network path is recorded, and a posterior settlement model is constructed by combining a preset network service quality evaluation mechanism.

In some embodiments of the present invention, a regular form game of the core cloud server and a cloud provider to which the edge cloud server belongs is constructed based on a game theory with the aim of maximizing own benefits, and the overall benefit calculation formula of the cloud provider is as follows:

where u (oi) represents the available revenue for the cloud provider; pi represents the likelihood of obtaining a corresponding benefit.

The invention has the advantages that:

the invention provides a distributed cloud-oriented wide area super-fusion service customization system, which constructs a distributed cloud-oriented wide area super-fusion service customization network architecture by starting from three planes of forwarding, control and service, and combines three technical layers of novel network bearing, a network operating system and cloud network super-fusion to realize deterministic transmission, application differentiation requirements and network service capacity used as required. By constructing a novel protocol stack layered structure centering on services, a wide area super fusion service customized network protocol stack facing the distributed cloud is designed, and the transmission control from TCP end to end is broken through from top to bottom by an application layer, so that the application-level, cross-wide area and distributed on-demand interconnection is realized. On the basis of application differentiation requirements and deterministic service capability, an interface between novel protocol stack layering is further designed, the network service capability is abstracted by introducing a concept of a network capability Ticket NaaS (network access service) in a service interconnection layer and a packet bearing layer, trust and anchoring relation between a network and an application is established, efficient cooperation between the network and the application and matching between a network path and the application are realized, and the relation between the network and the application is truly opened.

Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and drawings.

It will be appreciated by those skilled in the art that the objects and advantages that can be achieved with the present invention are not limited to the above-described specific ones, and that the above and other objects that can be achieved with the present invention will be more clearly understood from the following detailed description.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate and together with the description serve to explain the invention. In the drawings:

fig. 1 is a schematic diagram of a novel protocol stack layered structure of a wide area super fusion service customization system facing a distributed cloud according to an embodiment of the present invention.

FIG. 2 is a schematic diagram of a wide area super fusion service customization network architecture facing a distributed cloud according to an embodiment of the present invention.

Fig. 3 is a flow chart of a NaaS socket supply and demand matching mechanism with abstract service capability according to an embodiment of the invention.

Detailed Description

The present invention will be described in further detail with reference to the following embodiments and the accompanying drawings, in order to make the objects, technical solutions and advantages of the present invention more apparent. The exemplary embodiments of the present invention and the descriptions thereof are used herein to explain the present invention, but are not intended to limit the invention.

It should be noted here that, in order to avoid obscuring the present invention due to unnecessary details, only structures and/or processing steps closely related to the solution according to the present invention are shown in the drawings, while other details not greatly related to the present invention are omitted.

It should be emphasized that the term "comprises/comprising" when used herein is taken to specify the presence of stated features, elements, steps or components, but does not preclude the presence or addition of one or more other features, elements, steps or components.

It is also noted herein that the term "coupled" may refer to not only a direct connection, but also an indirect connection in which an intermediate is present, unless otherwise specified.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

It should be emphasized that the references to steps below are not intended to limit the order of the steps, but rather should be understood to mean that the steps may be performed in a different order than in the embodiments, or that several steps may be performed simultaneously.

The invention considers the great challenges facing the potential development trend of the prior art of merging service customized networks in the distributed cloud, and particularly has the problems of difficult to expect transmission quality, uncontrollable network and incapability of efficient cooperation between the network and the application. While constructing a cloud network technology facing the distributed cloud, various information and communication technology (Information and Communication Technology, ICT) services and resources (algorithms, programs, data, contents and the like) are intelligently and dynamically distributed and connected and cooperated as required at multiple levels of an operator 'cloud side end' even across multiple operators, and super-integration of network, calculation and storage is realized in a wide area.

The TCP/IP simplifies the open system interconnection (Open System Interconnection, OSI) model into a four-layer model, uses an IP protocol cluster as a network thin waist, and the technological gene of IP "best effort" results in the concomitant generation of TCP "end-to-end transmission control", and service semantics are not reflected in the whole hierarchy of TCP/IP, and meanwhile, the quality of service capability provided in a "patching" manner cannot effectively achieve deterministic transmission, differentiated requirements and target capability used as required by a service customization network. As shown in fig. 1, the present invention constructs a new protocol stack layered structure based on OSI seven-layer model, replacing the network layer with the service layer, redesigning each layer function, the new layered structure includes:

the application layer has the capabilities of an application layer, a presentation layer and a part of session layer in the OSI model and is mainly responsible for data encapsulation, encryption and decryption of data packets of each application.

And the service expression layer has the capability of a part of session layer and a part of transmission layer in the OSI model, and performs multidimensional service identification addressing on each application and attribute thereof of the application layer, wherein the attribute refers to the characteristic attached to the application in communication.

And the service transmission layer has the capability of part of the transmission layer in the OSI model, and adopts an open interface preset by a white box switch to control the transmission of the service customized network access side.

And the service interconnection layer has the capability of part of network layers in the OSI model, and adopts a software defined network controller (SDN controller) to realize interconnection and interworking of service customized networks.

And a packet carrying layer, which has the capability of a part of network layers and a part of data link layers in the OSI model, and uses a segment router to forward according to the network path appointed by each application data packet. Meanwhile, the forwarding capacity is packaged into a NaaS form, the forwarding capacity is matched with the requirements of each application, and a consumption matching mechanism is constructed.

And the data link layer has the capability of part of the data link layer in the OSI model, is responsible for adding a frame head and a frame tail at the head and the tail of a data packet respectively, sealing the frame, and adopting a time sensitive network switch to carry out end-to-end transmission on the data packet.

And the physical layer has the capability of the physical layer in the OSI model, adopts an Ethernet photoelectric interface to communicate, and provides a multidimensional resource pool constructed by a core cloud server and an edge cloud server, wherein the multidimensional resource pool at least comprises network resources, computing resources, storage resources and time resources.

Based on the novel layered structure provided by the invention, the system arrangement and redesign of the element functions of packet loss retransmission, order-preserving and duplication-removing, flow control, congestion control, flow engineering and the like in a data link layer, a network layer and a transmission layer are realized, the functions of a new data link layer and a new service transmission layer are strengthened, and the capability of customizing the systematic support network according to the requirement is supplied. Meanwhile, in order to realize 'customization on demand' of each application driver, the communication requirements among the applications are analyzed and processed, and the capability of each layer of the network is penetrated from top to bottom on the basis of functional layering.

In some embodiments, a systematic method is adopted at an application layer to analyze the communication requirements of each application, a requirement association matrix is generated, a representation paradigm of the communication requirements is constructed, and the communication requirements are expressed through a service expression layer.

Meanwhile, modeling a data flow, and performing byte coding mapping on a service transmission layer, a service interconnection layer and a data link layer by layer; the novel network coding theory (information theory) is adopted to realize the space-time control of network coding so as to drive the differential distribution of network space-time resources in a new physical layer.

As shown in fig. 2, on the basis of the above novel hierarchical structure, in order to realize the deterministic transmission of the service customization network, the existence of differentiated requirements and the target capability used as required, a framework of a wide area super fusion service customization system oriented to the distributed cloud is provided, which can be divided into three technical planes of novel network bearing, a network operating system and cloud network super fusion, and the forwarding plane, the control plane and the service plane of the network are systematically reconstructed, specifically, the novel network bearing technical plane corresponds to the forwarding plane, the network operating system technical plane corresponds to the control plane, and the cloud network super fusion technical plane corresponds to the service plane.

In the new network bearer technology plane, multiple domains are involved, and a new network data platform that is programmable, deterministic, customized and high-performance is built using segment routing routers, time-sensitive network switches (Time-Sensitive Network, TSNs) and flexible Ethernet (FlexE) optical interfaces. The method comprises the steps that a segment router is adopted to carry out programmable forwarding according to a network path appointed by each application data packet, and meanwhile reservation of network resources is realized; the time sensitive network switch is adopted to provide microsecond deterministic service for end-to-end transmission of data packets through the technologies of frequency clock synchronization, period mapping, flow shaping and the like; the Ethernet photoelectric interface is used for communication, so that a one-to-one mapping relation of strong binding of an original network layer and an original physical layer is broken, flexible and fine management of interface resources is realized, and meanwhile, the Ethernet photoelectric interface ensures high-pass bandwidth, elastic pooling and flexible packaged physical bottom performance and supports an optical signal PAMx modulation system.

The network operating system technical plane comprises a SDN (Software Defined Network) controller, and the SDN controller is an Operating System (OS) of the network, does not control network hardware, but operates as software, and is beneficial to network automation management. In the invention, the SDN controller is used as a centralized network brain, can collect the states of the service customized network in real time, including network topology, interface states, queue buffering and the like, and forms global dynamic load information of the service customized network.

Meanwhile, white box equipment is adopted to carry out data transmission based on an algorithm open interface, wherein white box equipment management is mainly based on a white box switch with software and hardware decoupled, bottom layer hardware and upper layer software can be customized according to the requirements of each application, compared with the traditional switch software and hardware binding purchase and monopoly use, the purchase cost of the switch can be obviously reduced, in addition, in the aspect of software functions, secondary development can be carried out based on open source software, and the development period and cost are reduced. In the invention, white box equipment management is adopted, forwarding logic of a data surface is customized in a software-defined mode, rapid development of various routing protocols is supported, and the agility, the manufacturability and the management of the network are unified and flexible.

Meanwhile, the network operating system performs unified abstract description modeling on each application, and the north interface light-weight report of the data surface is realized through the algorithm open interface design, so that the high-efficiency application under the requirement of a large-scale heterogeneous algorithm is supported. The network operation system is used as a technical plane for effectively connecting network resources of a forwarding plane and network requirements of a service plane, and a resource scheduling algorithm for user mode application (processes and threads) and user mode application (cooperative programs) in the computer operation system is used as a reference, so that transmission resource configuration required by application services (quintuple, connection and service/sub-service) is scheduled, the trust boundary problem between the two states is solved, and a universal intelligent extensible network management and control capability middle station is realized.

In the cloud network super-integration technical plane, the cloud network super-integration technical plane comprises a multi-level core cloud server and a multi-level edge cloud server, and a distributed cloud is formed by the core cloud server and the edge cloud server. The core cloud servers are located at core positions, and the edge cloud servers are close to the user side mobile terminal equipment and are subdivided into different layers according to distances from the user.

The cloud network super-fusion technology plane realizes wide area super-fusion integrated landscape of multi-dimensional service identification addressing, resource integrated multi-stage scheduling, large-scale multi-cloud exchange and other functions, and is specific:

carrying out multidimensional service identifier addressing on each application and attribute thereof in an application layer in a service expression layer, wherein the multidimensional service identifier addressing is divided into a basic identifier and a non-basic identifier; the attribute refers to a feature to which each application is attached at the time of communication. The basic identification is used for carrying out protogenic differentiation on different applications, calibrating service attributes, providing addressing capability of a two-dimensional service space, and the basic identification comprises identification of information such as programs, contents, applications and the like.

In some embodiments, when the basic identifier cannot cover part of the non-basic content, the non-basic identifier is formulated to identify each application by means of metadata, and exemplary non-basic identifiers include identifying information such as an IP address, a GPS, and the like.

In some embodiments, a hybrid scheduling framework is designed for cooperatively scheduling a plurality of heterogeneous resource clusters across regions on the basis of the multidimensional service identifier, so as to perform unified management and hierarchical scheduling on all nodes and/or clusters of the system. Wherein, hierarchical scheduling is embodied as follows: and screening according to the first-level scheduling index to obtain a target resource cluster, screening according to the second-level scheduling index to obtain a target node, and distributing corresponding application to the screened target node. Performing capacity segmentation between a first-level scheduling and a second-level scheduling, wherein the first-level scheduling index comprises geographic positions, use cost and cloud server information; the second level scheduling indicator includes a resource duty cycle, a power consumption, and a dependency association parameter between the resource and the corresponding node.

Meanwhile, the cloud network super-integration technical plane also builds a large-scale multi-cloud exchange interconnection system, and aims at the problems of inconsistent capability, functions and interfaces of each cloud platform such as public cloud, private cloud and industry cloud and heterogeneous cloud scenes, a distributed software service platform is designed, the overall stability and expansibility of the system are improved, and the management capability of the high-performance multi-cloud gateway is enhanced. Therefore, the cloud network super-fusion technology plane runs through computing and storage resources in the distributed cloud, and supports intelligent dynamic distribution of various application services in the distributed cloud, so that heterogeneous ubiquitous integrated cloud network business capability prototypes are realized.

Based on the three technical planes of the novel network bearing, the network operating system and the cloud network super-integration, the method can be further explored into a biplane form of a future network, the basic bearing network (comprising a forwarding plane and a control plane) is used as a first plane, and a second plane of the superposition service is formed by a distributed cloud service plane.

Macroscopically, the distributed cloud is composed of a core cloud server and an edge cloud server. The core cloud server is located at a plurality of core positions in the network, the network connectivity is high, the computing storage resources are rich, but the average distance from the mobile terminal at the user side is far, so that the access delay is increased. The edge cloud server is located at the edge position in the network, and is relatively low in network connectivity and relatively low in computing storage resources, but the average distance between the edge cloud server and the user side mobile terminal is relatively short, so that access delay is low. Core cloud servers of different levels can unload core cloud to form distributed remote of core cloud service capability, and can realize organic collaboration with the core cloud servers, and particularly, the core cloud servers can provide full service capability, and multistage edge cloud servers cooperatively realize dynamic caching of the service capability of the core cloud servers.

From the view point of the network, the edge cloud servers provide access control capability, the core cloud servers can provide flow optimization capability among the edge cloud servers, the structure can be independently arranged on an operator bearing network, the whole internet can be covered across different operators by utilizing rich network connectivity of the core cloud servers, and then a second plane of the internet is formed.

The underlying carrier network as a first plane may provide the capability for distributed inter-cloud interconnection of a second plane based on service customization ideas and network operating systems, including flexible scalable bandwidth and determination of predictable quality of service. The "overlay service plane" and the "base bearer plane" may be loosely coupled, but organically coordinated, evolution forms the biplane morphology of future networks.

In the prior art, the network and the application are relatively independent, and the realization of effective coordination still faces more problems. Trust and anchoring relation between the network and the application are technical keys for the development of a wide area super fusion service customized network facing the distributed cloud, and comprise two important mechanisms: firstly, network functions related to services are unhooked from the underlying network infrastructure through virtualization; secondly, network resources are encapsulated in Service entities and are exposed and utilized through abstract interfaces, and such a technology is called a Service-as-a-Service (NaaS).

Based on the principle, the invention encapsulates the differentiated requirements, deterministic transmission and other forwarding capacities of the service customization network into NaaS form in the packet bearing layer, introduces the concept of a network capacity Ticket (NaaS Ticket) to abstract the network service capacity, constructs a contractual model between passengers (services) and vehicles (bearing) through the concept of the Ticket, thereby establishing a set of consumption matching mechanism of the network service capacity, realizing the organic cooperation of the network capacity at the network side and the service requirement at the application side, and further truly opening the relationship between the network and the application.

As shown in fig. 3, the present invention constructs a NaaS socket supply and demand and matching mechanism with abstract service capability, and mainly includes: the difference determines an elastic network ticket supply mechanism, an application ticket consumption mechanism of on-demand statement posterior and a platform ticket matching mechanism of cross-domain timely and dynamic.

The method is oriented to the customizing requirement of communication among various applications, and mainly abstracts service capabilities such as network path delay, jitter, bandwidth and the like, builds a network ticket supply mechanism with difference determination elasticity, and comprises the following steps of S101 to S104:

step S101: and establishing an internal relation between the network space-time resource and the network path capacity by adopting a dimension analysis method.

Step S102: and adopting real-time online planning to realize differential scheduling of network paths under the condition of limited resources.

Step S103: packet time slot queuing is adopted to ensure the limitation of network path delay and jitter.

Step S104: based on single-path expansion and multi-path sharing, the flexible speed change of the network path broadband is realized.

In step S101, the service capabilities such as delay, jitter, bandwidth, etc. of the network path are essentially the demands for space-time resources such as forwarding queues, links, etc., so in order to establish an internal correlation between space-time resources of the network and the network path capabilities, a dimension analysis method is used to quantify the network capabilities, and a pricing model for resource consumption is constructed.

In step S102, the resource supply of the bearer network is limited, and in order to ensure differentiated requirements and efficient resource utilization, real-time online planning (operation study) is adopted to implement differentiated scheduling of network paths of resources under limited conditions.

In some embodiments, relevant factors affecting resource scheduling are abstracted into multidimensional vectors, the multidimensional vectors are subjected to orthogonal decomposition to obtain decision subsets with different dimensions, a multi-level decision tree is constructed, the whole decision process is completed cooperatively by decision submodules with different dimensions by utilizing the multi-level decision tree, and finally, a multi-level mixed distributed scheduling decision is generated, so that the communication requirements of each application are met. The relevant factors comprise resource duty ratio, power energy consumption, dependent association parameters between the resource and the corresponding nodes and resource energy consumption.

In step S103, time dimension resources, i.e., time slots, are reserved in advance for network traffic based on the resource abstraction of step S102 and the resource scheduling of step S103. To achieve deterministic quality of service, packet slot queuing (operations research) is employed to guarantee bounded network path delay and jitter.

In some embodiments, under the conditions of complex topology and mass flow, based on time slot constraint conditions, a real-time queuing model is built by taking jitter and time delay of a network path as targets, and a queuing scheme of each application data packet is generated by using the real-time queuing model, so that the problems of a queuing network and complex queuing are solved. The time slot constraint conditions comprise the size, the number, the packet sending time and time resources of the data packets. Meanwhile, according to the requirements of the deterministic service classification and classification, the deterministic classification can be realized through double-factor conversion of jitter and time delay.

In step S104, considering the network micro burst problem, that is, the network receives a plurality of burst data in units of milliseconds, so that the burst data rate is tens or hundreds of times higher than the average rate, even exceeds the port bandwidth, the network path bandwidth is elastically changed by adopting a method of mixing single-path expansion and multi-path sharing, and a proper amount of micro burst is supported by promise and tolerance dual-rate adjustment.

Referring to standard specification and end-to-end modeling of telecommunication standard department ITU-T for quality of service (Quality ofService, qoS), an application ticket consumption mechanism of a service customization network on-demand declaration posterior is constructed, and the specific flow comprises the following steps S201 to S204:

step S201: the inherent association between system throughput concurrency and inter-application communication end-to-end QoS is mined by adopting the distributed system theory.

Step S202: and the feedback control theory is adopted to realize the on-demand expression of the QoS from the application end to the end and the automatic decomposition of the network performance NP.

Step S203: and establishing a network service capability evaluation mechanism based on the mixing of probe dial testing and in-band telemetry.

Step S204: and realizing network path consumption accounting based on the actually-occurring traffic observation and service capability evaluation.

In step S201, in order to further analyze the effect of statistical quality of service QoS on system throughput, the inherent association between system throughput concurrency and inter-application communication end-to-end QoS is mined using distributed system theory (system theory). And in cooperation with the multi-cloud server distributed system, splitting the service into small-granularity micro-services which can be distributed and deployed on different independent machines, and establishing a budget model facing the system utility. Meanwhile, in order to ensure the stability of online service, the occurrence of an abnormal situation of a single cloud server is prevented, so that an avalanche effect is caused, and the high availability of the system is ensured by adopting a cluster mode.

In step S202, steady-state quantization of processing performance of each node is realized by adopting a feedback control theory (control theory), a network optimization model is constructed by pre-training a service demand budget model and collecting network state information obtained by analyzing feedback in real time, and an effective optimization mechanism for each data in a service customization network is learned so as to solve the problems of data transmission QoS guarantee, link load balancing and the like, optimize the service customization network, and realize on-demand expression of QoS from each application end to end and automatic decomposition of network performance NP. The network state information comprises link load, link bandwidth, link delay, link jitter, link packet loss rate, link utilization rate and the like.

In step S203, in order to realize that the network operating system technical plane can accurately sense the state of the network, a method of mixing probe dial-testing and in-band telemetry is adopted to observe the actual delay and jitter of the network path, and a method of injecting and extracting metadata into the production environment message is adopted to sense the network state information in real time. Meanwhile, based on high-precision network information and communication flow information active measurement, a network service capability QoS evaluation mechanism is constructed by declaring deviation variance.

In step S204, based on the observation of the actually generated traffic in step S203, a network actual consumption model is constructed according to the network status information obtained by real-time sensing, the consumption record of the network path is recorded, and a posterior settlement model is established in combination with the network service capability QoS evaluation mechanism obtained in step S203.

Combining a network ticket supply mechanism and an application ticket consumption mechanism, and referring to an end-to-end travel service concept of a travel traffic third party platform, constructing a cross-domain timely dynamic platform ticket matching mechanism, wherein the specific flow comprises the following steps of S301 to S304:

step S301: and adopting game theory to realize alliance credible co-treatment under the environment of multi-subject participation of the application network.

Step S302: and constructing a communication model which spans multiple autonomous network domains between applications based on the transfer cross-over idea.

Step S303: the application non-blocking type calling of the short stream is guaranteed based on a two-stage submitting mechanism.

Step S304: and the network service capability following after the application cross-domain migration is realized based on the consistency updating method.

In step S301, under a wide area super fusion service customization network architecture facing the distributed cloud, on-demand customization of service requirements often involves cloud providers to which a plurality of cloud servers belong, and adopts game theory (operation research) to solve the problem of the camping general in the multi-subject participation environment, and realizes trusted co-treatment of the alliance through dynamic bidding rational excitation. Under the multi-cloud provider system, each cloud provider acts as an agent, and each agent aims to maximize own benefits, so that a concept of regular Form Game (Normal-Form Game) is introduced, assuming that there are N agents in the system, service actions that each agent can do are contained in a, service benefits obtained by each agent are recorded as u, a triplet (N, a, u) is constructed, and a Utility Function (Utility Function) is constructed.

In some embodiments, the overall benefit calculation for all cloud providers is as shown in equation (1):

where u (oi) represents the revenue available to each cloud provider; pi represents the likelihood of obtaining a corresponding benefit.

The overall utility function may be broken up into several sub-utility functions, which in some embodiments are solved using decision trees. And furcation is carried out according to the value of each random variable, and finally, the utility function value on each leaf of the decision tree is obtained to obtain an optimal decision, so that the benefit maximization of each cloud provider is realized, and the on-demand customization capability of the service customization network is realized.

In step S302, traffic transmission across the wide area network needs to construct a communication model across multiple autonomous network domains between applications, and based on the transfer multiplication idea, that is, transmission needs to pass through another or several transfer stations from the origin to the destination, and the problem of time delay decomposition of the cross-domain network path is solved by integrating multiple factors such as credit cost and the like.

In step S303, in order to maintain the consistency of the information of the routing policies on the basis of all the nodes under the distributed system architecture, a two-stage commit mechanism algorithm is designed, specifically, a network operating system as a coordinator is introduced to uniformly govern the routing policies of all the nodes (hereinafter referred to as participants) and finally indicate whether or not the nodes are to actually commit the routing policies. The method comprises the steps that whether a participant realizes a routing strategy or not is informed to a coordinator, and then the coordinator decides whether each participant needs to continue transmission or stop transmission according to feedback information of all participants, so that data head packet retention during connection establishment is avoided, and timely application non-blocking calling of short flows is guaranteed.

In step S304, after steps S301 to S303, the following of the network service capability after the application cross-domain migration is realized based on the consistency updating method, so as to ensure the dynamic maintenance of the communication requirement between applications.

Meanwhile, in the novel layered structure provided by the invention, the design of the interface between the application layer and the service expression layer has obvious difference with the interface between the application layer and the transmission layer in the TCP/IP technical system. This is because each application is mainly represented as "reachable" with the interface semantics of the transport layer under the condition of using TCP/IP as the network interconnection basis, and what is needed in the service customization network provided by the present invention is "use on demand".

In some embodiments, the invention remodels the semantics of the network programming part in the application programming interface (Portable Operating System Interface of UNIX Application Programming Interface, POSIX API) of the traditional portable operating system, replaces the front end and the back end of the traditional Socket by using the API and/or using a software development kit (Software Development Kit, SDK), allows the application to actively express the requirement on the network, is similar to a process of buying a ticket, and finally obtains the required network capability by holding the ticket after paying according to the ticket price through the analytic mapping of the middle multiple stages.

The embodiments of the present invention also provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the edge computing server deployment method described above. The computer readable storage medium may be a tangible storage medium such as Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disk, a removable memory disk, a CD-ROM, or any other form of storage medium known in the art.

In summary, the invention provides a distributed cloud-oriented wide area super-fusion service customization system, which constructs a distributed cloud-oriented wide area super-fusion service customization network architecture by starting from three planes of forwarding, control and service, and combines three technical layers of novel network bearing, a network operating system and cloud network super-fusion to realize deterministic transmission, application differentiation requirements and network service capability used as required. By constructing a novel protocol stack layered structure centering on services, a wide area super fusion service customized network protocol stack facing the distributed cloud is designed, and the transmission control from TCP end to end is broken through from top to bottom by an application layer, so that the application-level, cross-wide area and distributed on-demand interconnection is realized. On the basis of application differentiation requirements and deterministic service capability, an interface between novel protocol stack layering is further designed, the network service capability is abstracted by introducing a concept of a network capability Ticket NaaS (network access service) in a service interconnection layer and a packet bearing layer, trust and anchoring relation between a network and an application is established, efficient cooperation between the network and the application and matching between a network path and the application are realized, and the relation between the network and the application is truly opened.

Those of ordinary skill in the art will appreciate that the various illustrative components, systems, and methods described in connection with the embodiments disclosed herein can be implemented as hardware, software, or a combination of both. The particular implementation is hardware or software dependent on the specific application of the solution and the design constraints. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention. When implemented in hardware, it may be, for example, an electronic circuit, an Application Specific Integrated Circuit (ASIC), suitable firmware, a plug-in, a function card, or the like. When implemented in software, the elements of the invention are the programs or code segments used to perform the required tasks. The program or code segments may be stored in a machine readable medium or transmitted over transmission media or communication links by a data signal carried in a carrier wave.

It should be understood that the invention is not limited to the particular arrangements and instrumentality described above and shown in the drawings. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the present invention are not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between steps, after appreciating the spirit of the present invention.

In this disclosure, features that are described and/or illustrated with respect to one embodiment may be used in the same way or in a similar way in one or more other embodiments and/or in combination with or instead of the features of the other embodiments.

The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, and various modifications and variations can be made to the embodiments of the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A distributed cloud-oriented wide area superset service customization system, comprising:

the application layer is used for carrying out data encapsulation, encryption and decryption on the data packets of each application;

the service expression layer is used for carrying out multidimensional service identification addressing on each application and the attribute thereof of the application layer, wherein the attribute refers to the characteristic attached to each application in communication;

the service transmission layer adopts an open interface preset by the white box switch to control the transmission of the service customized network access side;

The service interconnection layer adopts a software defined network controller to realize the connection communication between the service customization network and each wide area network;

the packet bearing layer adopts a segmented router to forward according to a network path of a designated data packet at each node of the system; the packet bearing layer encapsulates the forwarding capability into NaaS form, matches the forwarding capability with the requirements of each application, and constructs a consumption matching mechanism;

the data link layer is used for adding a frame head and a frame tail at the head and the tail of the data packet respectively, sealing the frame, and carrying out end-to-end transmission on the data packet by adopting a time sensitive network switch;

the physical layer adopts an Ethernet photoelectric interface for communication; the physical layer provides a multidimensional resource pool constructed by a core cloud server and an edge cloud server, wherein the multidimensional resource pool at least comprises network resources, computing resources, storage resources and time resources.

2. The distributed cloud-oriented wide area super fusion service customization system according to claim 1, wherein the service expression layer addresses multidimensional service identification for each application and its attributes of the application layer, further comprising:

On the basis of the multi-dimensional service identification, multi-dimensional resources in the multi-dimensional resource pool are subjected to hierarchical scheduling, a target resource cluster is obtained through screening according to a first-level scheduling index, a target node is obtained through screening according to a second-level scheduling index, and tasks of corresponding applications are distributed to the target node; the first-level scheduling index comprises geographic position, use cost and cloud server information; the second-level scheduling index comprises a resource duty ratio, power consumption and dependent association parameters between the resource and the corresponding node.

3. The distributed cloud oriented wide area super fusion service customization system according to claim 1, wherein the communication requirements of each application are analyzed at the application layer, a requirement correlation matrix is generated, a representation paradigm of the communication requirements is constructed, and the communication requirements are expressed through the service expression layer.

4. The distributed cloud-oriented wide area super fusion service customization system according to claim 1, wherein relevant factors affecting resource scheduling are abstracted into multidimensional vectors, orthogonal decomposition is carried out on the multidimensional vectors to obtain decision subsets with different dimensions, a multi-level decision tree is constructed, and scheduling decisions are generated by using the multi-level decision tree; the related factors include resource duty cycle, power consumption, dependent correlation parameters between the resource and the corresponding node, and resource energy consumption.

5. The distributed cloud-oriented wide area super fusion service customization system according to claim 1, wherein a real-time queuing model is constructed based on time slot constraint conditions and with the jitter and time delay of a network path as targets, and a queuing scheme corresponding to application data packets is generated by using the real-time queuing model; the time slot constraint conditions comprise data packet size, number, packet sending time and time resource.

6. The distributed cloud oriented wide area super fusion service customization system according to claim 1, wherein a single-path telescoping and multi-path sharing mixed method is adopted to flexibly change the network path bandwidth so as to receive burst data.

7. The distributed cloud oriented wide area super fusion service customization system according to claim 1, wherein network state information obtained by analysis feedback is collected in real time, a network optimization model is built, and the service customization network is optimized; the network state information includes link load, link bandwidth, link delay, link jitter, link packet loss rate, and link utilization.

8. The distributed cloud oriented wide area super fusion service customization system according to claim 7, wherein a probe dial testing and in-band telemetry hybrid method is adopted to observe the actual delay and jitter of a network path, and metadata is injected and extracted into a production environment message to sense the network state information in real time.

9. The distributed cloud oriented wide area super fusion service customization system according to claim 8, wherein a network actual consumption model is built according to the network state information obtained through real-time sensing, consumption records of network paths are recorded, and a posterior settlement model is built by combining a preset network service quality evaluation mechanism.

10. The distributed cloud-oriented wide area super fusion service customization system according to claim 1, wherein regular form games of the core cloud server and cloud providers to which the edge cloud server belongs are built based on game theory with the aim of maximizing own benefits, and the overall benefit calculation formula of the cloud providers is as follows:

where u (oi) represents the revenue available to the cloud provider; pi represents the likelihood of obtaining a corresponding benefit.

Images