WO2012065521A1

WO2012065521A1 - Hierarchical distributed cloud computing system and method for providing service

Info

Publication number: WO2012065521A1
Application number: PCT/CN2011/081965
Authority: WO
Inventors: 沈玉龙; 裴庆祺; 张志为; 习宁; 尹浩; 徐启建; 马建峰; 姜晓鸿; 吴作顺; 唐宏
Original assignee: 西安电子科技大学; 中国电子设备系统工程公司研究所
Priority date: 2010-11-16
Filing date: 2011-11-09
Publication date: 2012-05-24
Also published as: CN101977242A

Abstract

Disclosed are a hierarchical distributed cloud computing system and a method for providing service. The system is composed of a core cloud, edge clouds and terminals. The core cloud is a cloud computing service centre composed of servers with powerful storage and computing capabilities; an edge cloud, connected to the core cloud via a backbone network, is composed of server nodes distributed in different networks and areas; and the terminal is an entity using the cloud computing service provided by the core cloud and the edge cloud. The core cloud and the edge cloud cooperate to provide cloud computing service to the end user and process the data traffic by using the relationship between the core cloud and the data transmitted by the terminal, reducing communication traffic. The present invention confines the majority of network traffic in the cloud computing in a local network between the edge cloud and the terminal, reducing the traffic in the backbone network, effectively avoiding problems such as network congestion, high latency and so forth which exist in the current centralized cloud computing model, and improving user experience.

Description

Hierarchical distributed cloud computing system and service providing method

The present invention relates to the field of computers, and in particular to a layered distributed cloud computing system and a service providing method. By introducing the concept of an edge cloud, the reliance of cloud computing on network bandwidth is reduced, and the QoS of the cloud computing is improved (Quality of Service, quality of service). Background technique

Cloud Computing is a more reasonable and effective service provision and operation mode in computer networks. In this mode, traditionally, various local computing, storage and other services are supported by large-scale, centralized cloud computing services. Replace. In cloud computing, users can easily and transparently use various computing, storage and other resources provided by cloud computing service providers on demand. These computing resources can be quickly provisioned and released, reducing the cost of acquisition, management, and maintenance for enterprise or individual users, with broad application prospects. Cloud computing is a service operation mode proposed by the computer business industry. The related technologies involve Grid Computing, Distributed Computing, Parallel Computing, Utility Computing, and Network Storage. Network Storage Technologies ), Virtualization, Load Balance, and more.

One of the core concepts of cloud computing is to continuously improve the processing power of the "cloud" end (computing center), thereby reducing the processing load on the user terminal, and finally simplifying the user terminal into a simple input and output device, and enjoying the "cloud" on demand. "The powerful computing power of processing."

Under the existing centralized cloud computing architecture, a centralized large-scale data center and computing center are established to form a cloud computing cloud, and the cloud provides external services required by users; users obtain services directly from the cloud through the network. , As shown in Figure 1. Since the user's data storage and calculation are performed in the cloud, cloud computing has a strong dependence on the bandwidth and data transmission delay of the network. The application of cloud computing inevitably leads to a large amount of data interaction, which puts heavy pressure on the export network of the computing center. The traffic of the cloud egress network is the sum of the traffic of each terminal and the cloud. In the current network environment, the quality of service QoS (availability, throughput, latency, delay variation, etc.) that a centralized cloud computing architecture can provide is difficult to meet the user's demand for cloud computing; for example, remote virtual desktop applications are subject to Network congestion, high latency and other problems. In order to guarantee the QoS of cloud computing, various related technologies are studied, the most typical of which is to compress data to reduce the amount of data transmitted on the network and reduce the urgent need for cloud computing to network bandwidth. However, these technologies can only partially alleviate the network burden in the cloud computing environment, and do not fundamentally solve this problem. Summary of the invention

The problem to be solved by the present invention is that the existing computer network is difficult to meet the requirements of cloud computing services in terms of bandwidth, delay, and the like. In order to solve this problem, the existing cloud computing model is improved, and a layered and distributed cloud computing system is designed. The corresponding server nodes are set up to form an edge cloud, which is composed of "core cloud", "edge cloud" and user terminals. Hierarchical distributed cloud computing system. The edge cloud cooperates with the core cloud to process the correlation between communication data, reduce the amount of data exported by the core cloud, and solve the problem of cloud computing dependence on network bandwidth and delay.

The solution to the technical problem of the present invention is that the hierarchical distributed cloud computing system is composed of a core cloud, an edge cloud and an end user:

Core cloud: It has powerful computing and storage capabilities, and cooperates with the edge cloud to provide users with transparent cloud computing services quickly and flexibly. The core cloud is a cloud computing service center consisting of a cluster of servers with powerful storage and computing capabilities. Users use the various services provided by the core cloud according to their needs.

Edge Cloud: It consists of server nodes distributed in different networks and regions, providing corresponding auxiliary functions for the core cloud, and collaborating with the core cloud to provide cloud computing services to users quickly and flexibly. The edge cloud is connected to the core cloud through the backbone network, and the user terminal uses the services provided by the edge cloud in the vicinity of the core cloud. On the one hand, the edge cloud is responsible for processing the data stream between the core cloud and the user terminal (such as compression, encryption and decryption, etc.), using the correlation between communication data, reducing network overhead, reducing delay, and ensuring cloud computing service quality. QoS; On the other hand, the edge cloud storage terminal accesses the common data and common data required by the cloud computing service (for example, files frequently accessed by users, graphics libraries required for generating graphical user interfaces, etc.).

Terminal: A requesting entity that uses a cloud computing related service. The main responsibility of the end user is to complete the input and output functions: On one hand, log in to the cloud computing center, build the user work environment, request the required service from the cloud computing; on the other hand, provide the user with an interface for browsing and displaying the cloud computing service.

In a hierarchical and distributed cloud computing system, the core cloud and the edge cloud cooperate to provide users with cloud computing services that guarantee QoS:

Collaboration between the edge cloud and the core cloud: The core cloud provides cloud computing services such as user computing and storage; the edge cloud shares the related functions of the core cloud, and the common data and user common data required for the storage terminal to access the cloud computing service are coordinated in the core cloud. Under the transparent, transparently provide users with cloud computing services. The high-bandwidth and low-latency network features between the edge cloud and the terminal ensure the QoS of the cloud computing service. The core cloud and the edge cloud collaborate to provide cloud computing services to end users in a transparent manner to users;

Collaboration between the edge cloud and the terminal: The edge cloud uses the high-bandwidth and low-latency local network between the edge cloud and the terminal to quickly transmit the common data and user common data required for the terminal to access the cloud computing service to the terminal, and transparently to the terminal. Offer Real-time cloud computing services;

Collaboration between the core cloud and the terminal: With the aid of the edge cloud, the core cloud completes the specific computing and storage services requested by the terminal.

The service provision method based on the hierarchical distributed cloud computing system is that the core cloud and the edge cloud cooperate together to transparently provide cloud computing services for users. The user first makes a service request to the core cloud, and then the core cloud negotiates the service policy with the edge cloud where the user is located, and finally cooperates to provide the user with the QoS guarantee service according to the result of the negotiation.

In the cloud computing system of the present invention, according to different scenarios, the data processing is used to provide the data processing process for the core cloud to the terminal, that is, the core cloud and the edge cloud negotiation service policy where the user is located, and finally according to the negotiation result. , work together to provide users with guaranteed QoS services, there are three ways:

Method 1: For the case where the data M to be sent to a certain terminal of the core cloud has its own correlation, the core cloud negotiates with the edge cloud server where the target terminal is located according to the characteristics of the data M, and processes the data M according to the result of the negotiation. The data m with smaller data volume is generated, and the data m is sent to the edge cloud server; after receiving the data m, the edge cloud server uses the reverse algorithm to obtain the original data M, and sends the original data M to the target terminal. The terminal sends data to the core cloud, taking the opposite process.

Method 2: For providing data with correlations for multiple terminals, set the data to Μ^Μ^. Μ^ The core cloud negotiates with the edge cloud, and uses the result of the negotiation to perform data M^M^. J^ Processing, generating a data set with a smaller amount of data, and sending it to the edge cloud server; after receiving the data set, the edge cloud server restores the data set, and sends the restored data to the corresponding terminal. The terminal sends data to the core cloud, taking the opposite process.

Method 3: The third case is that the target data to be sent by the core cloud is Μ^Μ^. , Μ^ The process of generating the target data is the same, and there is a common input (for example, using the same graphics library); therefore, the target data not directly generated by the core clouds, but under the coordination of cloud core, the core edge cloud cloud negotiation, according to a result of negotiation, the MM _2, ... M _n corresponding to the target command and data sent to the data terminal After the edge cloud server receives the command, the edge cloud server calls the negotiated correlation function to calculate the target data according to the relevant data, and sends the target data to the corresponding terminal. For example, in a remote desktop application, the generation of graphical user desktops is moved from the core cloud to the edge cloud; a large amount of graphics data is only transmitted between the edge cloud and the terminal connected to it, thereby reducing network transmission on the core cloud outlet. The amount of data.

When the location of the end user changes, the data associated with the user needs to be migrated to better meet the user experience. The migration is divided into live migration and planned migration depending on whether the user is using the cloud computing service when the location changes.

Real-time migration refers to the location movement of the user in the process of using the cloud computing service, and the edge cloud is moved into the new edge cloud. At this time, the user service data should be migrated from the original edge cloud to the new edge cloud under the coordination of the core cloud. New at this time The edge cloud replaces the original edge cloud to provide services to users.

Planned migration means that the user informs the cloud in advance of his or her schedule. The cloud computing center plans to arrange user-related data (or user service data) between different edge clouds according to the user's itinerary, and the user service data is located. The edge cloud provides services to users, ensuring that users can use the services provided by cloud computing in real time during the scheduled itinerary.

The invention limits the network traffic in most cloud computing to the end user local, reduces the traffic of the backbone network, effectively avoids the dependence of the centralized cloud computing mode on the network bandwidth, the delay, etc., and ensures the QoS of the cloud computing. .

The invention has the beneficial effects compared with the prior art:

The layered and distributed cloud computing architecture designed by the invention solves the problem that the existing network bandwidth and delay cannot meet the requirements of the cloud computing service from the perspectives of the network and the storage location. The invention sets up an edge cloud composed of server nodes with computing and storage capabilities in various networks and regions, and forms a hierarchical distributed cloud computing architecture with three levels of "core cloud", "edge cloud" and user terminal. The nodes in the edge cloud use the correlation between communication data by cooperating with the core cloud (the correlation between communication data includes the correlation between its own data and the correlation between parallel data, through network coding, compression technology, automatic caching). Technologies such as the local graphics library can effectively reduce the amount of data.) Reduce the amount of data exported by the core cloud. Therefore, a hierarchical and distributed cloud computing architecture is adopted, and user desktop environment data and user common data are stored on a server closer to the user. This method can limit the data transmission of cloud computing services to users locally, reduce the data traffic of the backbone network, fundamentally solve the dependence of cloud computing on network bandwidth and delay, and ensure the availability of cloud computing. DRAWINGS

Figure 1 is a schematic diagram of an existing centralized cloud computing network system;

2 is a schematic diagram of a hierarchical distributed cloud computing system proposed by the present invention;

3 is a schematic diagram of a service providing method for hierarchical distributed cloud computing proposed by the present invention;

Figure 4 is a schematic diagram of traffic under the existing centralized cloud computing system;

FIG. 5 is a schematic diagram of reducing traffic by utilizing correlation between self data according to the present invention; FIG.

6 is a schematic diagram of reducing traffic according to correlation between parallel data proposed by the present invention;

7 is a schematic diagram of the present invention for moving a key computation causing data expansion to an edge cloud to reduce traffic; FIG. 8 is a schematic diagram of a core cloud and edge cloud negotiation process proposed by the present invention;

9 is a schematic diagram of a real-time migration interaction process proposed by the present invention;

FIG. 10 is a schematic diagram of the interaction process of the planned migration parties proposed by the present invention. detailed description

The invention sets an edge cloud composed of corresponding server nodes in the cloud, and forms a hierarchical distributed cloud computing system (or hierarchical distributed cloud computing architecture) composed of "core cloud", "edge cloud" and user terminals. , as shown in picture 2. By setting up compute nodes between the core cloud computing center and the terminal to reduce traffic at the core cloud exit, the latency is reduced and the QoS of the cloud computing is guaranteed.

Core cloud: It has powerful computing and storage capabilities, and cooperates with the edge cloud to provide users with transparent cloud computing services quickly and flexibly. The core cloud is a cloud computing service center consisting of a cluster of servers with powerful storage and computing capabilities. Users use the services provided by the core cloud according to their needs. The servers that make up the core cloud can be called core cloud servers.

Edge Cloud: It consists of server nodes distributed in different networks and regions, providing corresponding auxiliary functions for the core cloud, and collaborating with the core cloud to provide cloud computing services to users quickly and flexibly. The edge cloud is connected to the core cloud through the backbone network, and the user terminal uses the services provided by the edge cloud in the vicinity of the core cloud. On the one hand, the edge cloud is responsible for processing the data stream between the core cloud and the user terminal (such as compression, encryption and decryption, etc.), using the correlation between communication data, reducing network overhead, reducing delay, and ensuring cloud computing QoS; On the one hand, the edge cloud storage terminal accesses common data and common data required by the cloud computing service (for example, files frequently accessed by users, graphics libraries required for generating a graphical user interface, etc.). The servers that make up the edge cloud can be called edge cloud servers.

Terminal: A requesting entity that uses a cloud computing related service. The main responsibility of the end user is to complete the input and output functions: On the one hand, log in to the cloud computing center, build the user work environment, request the required service from the cloud computing; on the other hand, provide the user with an interface for browsing and displaying the cloud computing service.

On the one hand, for the case that the data M to be sent to a certain terminal of the core cloud has its own correlation, the core cloud server is used to negotiate with the edge cloud server where the target terminal is located according to the characteristics of the data M, and according to the result of the negotiation, the data is M processes, generates data m with smaller data volume, and sends data m to the edge cloud server; the edge cloud server, after receiving the data m, uses the inverse algorithm to restore the original data M, and the original data M Send to the target terminal.

On the other hand, for the core cloud to provide relevant data to multiple terminals, set the data to

MM ₂ ,...M _n , the core cloud server, used to negotiate with the edge cloud, use the result of the negotiation, process the data ^, ₂ , .. „, generate a data set with a smaller amount of data, and send it The edge cloud server is used to receive the data set, restore it, and send the restored data to the corresponding terminal.

On the other hand, corresponding to the case where the target data ^^...^ is generated in the same process and there is a common input, the core cloud server is used to find that the target data M^M^. J^ is generated in the same process, and there is a common input. When negotiating with the edge cloud, according to the result of the negotiation, the command corresponding to the target data M^M^.J^ and related data are sent to the end The edge cloud server where the terminal is located; the edge cloud server, after receiving the command, calls the negotiated correlation function to calculate the target data according to the relevant data, and sends the target data to the corresponding terminal.

Further, the core cloud server is further configured to: when the terminal moves in the process of using the cloud computing service, when the current edge cloud moves into the new edge cloud, the user service data corresponding to the terminal is migrated from the current edge cloud to the new one. The edge cloud, and the new edge cloud provides services to users; or, the core cloud server is also used to pre-acquire the user's itinerary, and the user service data is transferred between different edge clouds according to the user's itinerary, and is served by the user. The edge cloud where the data resides provides services to users.

Under the layered distributed architecture, the core cloud and the edge cloud work together to transparently provide cloud computing services to users, as shown in Figure 3. The steps for layered distributed cloud computing services are as follows:

Step a. The user sends a service request to the core cloud;

Step b. The core cloud finds the edge cloud where the user is located, and negotiates the service policy with the edge cloud where the user is located; Step c. According to the negotiated service policy, the core cloud uses the response data, and the edge cloud uses the stored data (the terminal accesses the cloud) The common data required by the computing service and the user commonly used data) provide a transparent service for the user.

In the cloud computing service provision, the data transmission mode under the traditional centralized cloud computing architecture is shown in Figure 4. The core cloud directly transmits the data M required by each terminal to each terminal. The backbone network needs to bear the transmission of all data, which brings a large burden to the network.

Under the layered distributed cloud computing architecture, according to different scenarios, the core cloud is used to provide data M for the terminal to process according to the data correlation. According to the characteristics of the data itself, the processing has the following three methods:

Method 1: For the case where the core M is to be sent to a certain terminal, the data M has its own relevance. The core cloud negotiates the service policy with the edge cloud server connected to the target terminal according to the characteristics of the data M, and uses the result of the negotiation, the core cloud and the edge. The cloud provides services to users together, as shown in Figure 5. The specific implementation steps are as follows:

Step a: The core cloud finds that the target data sent to each terminal has its own relevance;

Step b: The core cloud initiates a collaboration request to the edge cloud where the terminal is located;

Step c: After receiving the collaboration request, the edge cloud detects whether the resource meets the requirement, and if it meets the return acceptance response; otherwise, returns a rejection response;

Step d: If the core cloud receives the rejection, search for the new available edge cloud, and find the return step b, otherwise it fails; Step e: The core cloud receives the acceptance response, initiates the negotiation process with the edge cloud; negotiates the used data. Processing the algorithm, completing the negotiation process, and successfully establishing a collaborative relationship;

Step f: The core cloud processes the data M to be sent / (Μ,·)=«3⁄4·, generates the whole data m, and sends it to the edge cloud; after the edge cloud receives the data sent by the core cloud, it reverses Processing, that is, g(m)=M restores the data and sends it to the corresponding terminal terminal. Method 2: For the case of providing relevant data to multiple terminals, it is possible to set the data to M^M^. J^, the core cloud negotiates with the edge cloud, and uses the result of the negotiation, the data M^M^. J^ Processing, generating a data set with a smaller amount of data, and sending it to the edge cloud server; after receiving the data set, the edge cloud server restores the data set, and sends the restored data to the corresponding terminal. The terminal sends data to the core cloud, taking the opposite process. The data transmission process is shown in Figure 6. The specific implementation steps are as follows:

Step a: The core cloud finds a high correlation between the data sent to different terminals;

Step b: The core cloud initiates a collaboration request to the edge cloud;

Step f: The core cloud processes the data to be sent for each terminal together (MM.Mn Mw, ie u(M)=msg, generates the overall data Μ^, and sends it to the edge cloud; the edge cloud receives the core cloud to send After the data, it is inversely processed, that is, VJ = Μ decomposes the data component to be transmitted to each terminal, and transmits it to the corresponding terminal terminal.

Method 3: The third case is the target data to be sent by the core cloud.

The process of generating target data is similar or identical, and there is a common input (for example, using the same graphics library); therefore, the target data is no longer generated directly by the core cloud, but under the coordination of the core cloud, by the edge cloud The data is sent to the corresponding terminal. For example, in a remote desktop application, the generation of graphical user desktops is moved from the core cloud to the edge cloud; a large amount of graphics data is only transmitted between the edge cloud and the terminal connected to it, thereby reducing network transmission on the core cloud outlet. The amount of data. The data transmission process is shown in Figure 7. The specific implementation steps are as follows:

Step a: The core cloud discovers that the generation process of the target data sent to the multiple terminals is similar, and there is a common input; Step b: The core cloud initiates a collaboration request to the edge cloud;

Step d: If the core cloud receives the rejection, search for the new available edge cloud, find the return step b, otherwise fail; Step e: The core cloud receives the acceptance, initiates the negotiation process with the edge cloud; the core cloud will command semantics The negotiation processing algorithm is sent to the edge cloud, and the negotiation process is completed, and the collaboration relationship is successfully established;

Step f: When the core cloud wants to send data Mi to a terminal, the command for generating the data and the related small amount of data are sent to the edge cloud where the terminal is located;

Step g: After receiving the command and data, the edge cloud calls the relevant function, calculates the result, and sends the result to the phase. Close the terminal.

In the process of providing cloud computing services, the negotiation process of core cloud and edge cloud is shown in Figure 8. The specific negotiation steps are as follows:

Step a: After the user sends a service request, the core cloud determines the edge cloud according to the user location;

Step b: The core cloud initiates a collaboration request to the edge cloud;

Step c: After receiving the request, the edge cloud allocates resources for collaboration and sends a consent cooperation response to the core cloud. Step d: After receiving the consent response, the core cloud interacts with the edge cloud to negotiate the task assignment;

Step e: The core cloud and the edge cloud collaborate according to the negotiation results and provide services for users. When the location of the end user changes, the data related to the user (or user service data) needs to be migrated to better meet the user experience. The migration is divided into live migration and planned migration depending on whether the user is using the cloud computing service when the location changes.

In real-time migration, the user moves from the edge cloud to the new edge cloud. At this time, the user service data should be migrated from the original edge cloud to the new edge cloud under the coordination of the core cloud. At this point, the new edge cloud replaces the original edge cloud to provide services to users. The implementation steps for live migration are shown in Figure 9:

Step a: The terminal user sends the change and is perceived by the cloud;

Step b: The edge cloud determines whether it can still provide the required service according to the change information of the user's location; Step c: If the edge cloud determines that it can still provide the required service, repeat step b; otherwise, issue a live migration to the core cloud. Request

Step d: The core cloud selects a new available edge cloud based on the information provided by the edge cloud;

Step e: The core cloud negotiates with the new edge cloud and configures the required environment for the user;

Step f: The user work environment data is migrated from the original edge cloud to the new edge cloud;

Step g: The core cloud and the new edge cloud work together to provide cloud computing services to users.

In the planned migration, the cloud computing center schedules the transfer of user-related data between different edge clouds according to the user's itinerary, ensuring that users can use the services provided by cloud computing in real time during the scheduled itinerary. The implementation steps for the planned migration are shown in Figure 10:

Step a: The user submits his or her itinerary to the cloud;

Step b: The core cloud formulates a data transfer plan according to the user's itinerary;

Step c: The core cloud serves the user based on the data migration plan and with the corresponding edge cloud.

The embodiment of the invention designs a hierarchical and distributed cloud computing system, and solves the problem that the existing network bandwidth, delay and the like cannot meet the requirements of the cloud computing service from the perspectives of the network and the storage location. Set up calculations in each network and region, The server nodes of storage capacity form an edge cloud, forming a hierarchical distributed cloud computing architecture with three levels of "core cloud", "edge cloud" and user terminal. The nodes in the edge cloud use the correlation between communication data by cooperating with the core cloud (the correlation between communication data includes the correlation between its own data and the correlation between parallel data, through network coding, compression technology, automatic caching). Technologies such as the local graphics library can effectively reduce the amount of data.) Reduce the amount of data exported by the core cloud. Therefore, the hierarchical and distributed cloud computing architecture is adopted, and the user desktop environment data and the user common data are stored on a server closer to the user. This method can limit the data transmission of the cloud computing service to the user local, reduce the data traffic of the backbone network, fundamentally solve the dependence of the cloud computing on the network bandwidth and the delay, and ensure the availability of the cloud computing.

Claims

Claim

A hierarchical distributed cloud computing system, characterized in that: the system is composed of a core cloud, an edge cloud and a terminal; the core cloud is a cloud computing service center composed of servers with powerful storage and computing capabilities; The edge cloud is composed of servers distributed in different networks and regions, and is connected to the core cloud through the backbone network to provide corresponding auxiliary functions for the core cloud, and cooperate to provide cloud computing services to the users; the terminal uses the core cloud and the edge cloud. The entity providing the cloud computing service; the core cloud and the edge cloud negotiation service policy where the terminal is located, according to the result of the negotiation, work together to provide the terminal with the service of guaranteeing the quality of service QoS.

2. The hierarchical distributed cloud computing system according to claim 1, wherein:

The core cloud is used to cooperate with the edge cloud to ensure the quality of service (QoS) of the cloud computing service, and to provide cloud computing services to users who use the terminal in a transparent manner to the user;

The edge cloud is used for cooperation with the terminal, and uses the high-bandwidth and low-latency local network between the edge cloud and the terminal to quickly transmit the common data and user common data required for the terminal to access the cloud computing service to the terminal, and transparently The terminal provides real-time cloud computing services;

The core cloud is used for collaboration with the terminal. With the aid of the edge cloud, the core cloud completes the specific cloud computing service requested by the terminal.

3. The hierarchical distributed cloud computing system according to claim 1, wherein: the data M to be sent to a terminal by the core cloud has its own correlation,

The core cloud server is configured to negotiate with the edge cloud server where the target terminal is located according to the characteristics of the data M, process the data M according to the result of the negotiation, generate data m with smaller data volume, and send the data m to the edge cloud server. The edge cloud server, after receiving the data m, uses the inverse algorithm to restore the original data M, and sends the original data M to the target terminal.

4. The hierarchical distributed cloud computing system according to claim 1, wherein: for the core cloud, a plurality of terminals are provided with relevant data, and the data is M^M^^M^

The core cloud server is used to negotiate with the edge cloud, and uses the result of the negotiation to process the data M^M^.J^ to generate a data set with a smaller data amount and send it to the edge cloud server;

The edge cloud server, after receiving the data set, restores it, and sends the restored data to the corresponding terminal.

5. The hierarchical distributed cloud computing system according to claim 1, wherein:

Core cloud server, used to discover the target data M^M^. J^ is generated in the same process, and there is a common input When negotiating with the edge cloud, according to the result of the negotiation, the command and related data of the target data M^M^^M are sent to the edge cloud server where the terminal is located;

The edge cloud server, after receiving the command, calls the negotiated correlation function according to the related data to calculate the target data, and sends the target data to the corresponding terminal.

6. The hierarchical distributed cloud computing system of claim 1 wherein:

The core cloud server is also used to migrate the user service data corresponding to the terminal from the current edge cloud to the new edge when the terminal moves in the process of using the cloud computing service and moves from the current edge cloud to the new edge cloud. Cloud, and the new edge cloud provides services to users;

Or,

The core cloud server is also used to pre-acquire the user's itinerary, arrange user service data to be transferred between different edge clouds according to the user's itinerary, and provide services to the user by the edge cloud where the user service data is located.

A service providing method based on a hierarchical distributed cloud computing system, wherein: the service providing method comprises: a user submitting a service request to a core cloud, and a core cloud and an edge cloud negotiation service policy where the user is located, according to a result of the negotiation Work together to provide users with a service that guarantees quality of service QoS.

The tiered distributed cloud computing system-based service providing method according to claim 7, wherein: the core cloud and the edge cloud negotiation service policy where the user is located, according to the result of the negotiation, work together to provide the user with Services that guarantee quality of service QoS, including:

For the case where the data M to be sent to a certain terminal of the core cloud has its own correlation, the core cloud negotiates with the edge cloud server where the target terminal is located according to the characteristics of the data M, and processes the data M according to the result of the negotiation to generate the data amount. Smaller data m, and the data m is sent to the edge cloud server; after receiving the data m, the edge cloud server uses the inverse algorithm to restore the original data M, and sends the original data M to the target terminal;

Or,

For the core cloud to provide relevant data to multiple terminals, set the data to ^ ^... , the core cloud negotiates with the edge cloud, and uses the result of the negotiation to process the data M^M^ . J^ to generate the data volume. a smaller data set is sent to the edge cloud server; after receiving the data set, the edge cloud server restores the data set, and sends the restored data to the corresponding terminal;

Or,

The core cloud discovery target data M^M^. The generation process of J^ is similar, and there is a common input; the core cloud negotiates with the edge cloud, and according to the result of the negotiation, the target data M^M^. J^ corresponds to the command and related The data is sent to the terminal where it is located. After receiving the command, the edge cloud server calls the negotiated correlation function to calculate the target data according to the relevant data, and sends the target data to the corresponding terminal.

9. The method according to claim 7, wherein the method further comprises:

When the terminal moves in the process of using the cloud computing service and moves from the current edge cloud to the new edge cloud, the core cloud migrates the user service data corresponding to the terminal from the current edge cloud to the new edge cloud, and is new The edge cloud provides services to users;

Or,

The core cloud pre-acquires the user's itinerary, arranges the user service data to be transferred between different edge clouds according to the user's itinerary, and provides services to the user by the edge cloud where the user service data is located.