Cloud computing API translation system

Abstract

The present invention provides an API interface translation system for cloud computing, comprising: an interface management system, an interface arrangement system and an interface access management system; the interface management system is used for unified management of interfaces of various platforms, realizing the addition and deletion of interfaces, and managing interface related information; the interface arrangement system is used for automatically or manually arranging interfaces under the demand of application development, generating routes for interface access after automatically or manually selecting required interfaces, and unifying all selected interfaces into an externally exposed interface according to the transmission protocol required by the application, which is handed over to the interface access management system for management; the interface access management system is used for centralized management, monitoring and access optimization of the arranged interfaces, and can arrange interfaces according to the different functions realized for different platforms, expose interfaces on demand, convert interface data protocols, unify data protocols, and reduce the difficulty of application development.

Description

Technical Field

[0001] The present invention relates to the field of cloud computing, and in particular to an API interface translation system for cloud computing. Background Art

[0002] At present, the development of an application needs to interact with multiple backend service interfaces. This approach has a simple structure. However, since the functions provided by a single backend service are relatively simple, an application may need to initiate multiple requests to query a piece of information, resulting in high network overhead. The data protocols used for different interfaces are not unified, and the application needs to support multiple protocols, which increases the development workload. It also brings challenges to the subsequent business reconstruction. Summary of the invention

[0003] The present invention provides an API interface translation system for cloud computing to solve at least one of the above technical problems.

[0004] To solve the above problems, as one aspect of the present invention, a cloud computing API interface translation system is provided, comprising: an interface management system, an interface orchestration system and an interface access management system;

[0005] The interface management system is used to uniformly manage the interfaces of each platform, realize the addition and deletion of interfaces, and manage the interface related information;

[0006] The interface arrangement system is used to automatically or manually arrange interfaces according to the needs of application development, generate routes for interface access after automatically or manually selecting the required interfaces, and unify all selected interfaces into an externally exposed interface according to the transmission protocol (such as HTTPS) required by the application and hand it over to the interface access management system for management;

[0007] The interface access management system is used to centrally manage, monitor and optimize access to the orchestrated interfaces. After processing the received request, it sends requests to various interfaces in the background, encapsulates the data returned by the background, and returns it to the application through a unified transmission protocol.

[0008] Preferably, in the interface management system:

[0009] Define each platform interface as Interface(I1,I2,…In;Type), where In represents the number of interfaces included in the platform and Type represents the type of the platform;

[0010] The sub-interfaces of the platform are defined as I(A(c), A(p), Desr, Tans, Tag), where A(c) represents the abstract front-end request service list, A(p) represents the abstract back-end response service list, Desr represents different interface descriptions, Tans represents the transmission connection method, and Tag represents the abstract interface identifier.

[0011] Preferably, in the automatic arrangement mode, according to a series of interfaces Interfacei (I1, I2, ... In) required for application development, and defining the services A (c) required by each interface, and the transmission connection mode Tans, the similarity Sim (Ii, Ij) between each interface required by the application and the interface existing in the system is calculated in the interface management system, and a one-to-one correspondence is performed according to the maximum similarity matching principle. The similarity is calculated as follows:

[0012]

[0013] in,

[0014] Among them, Parameter represents the interface parameter, and the above formula represents the comparison and calculation of various parameters of the interface Ii required for application development and the interface Ij in the interface management system.

[0015] Preferably, in the manual orchestration mode, the required interfaces are selected in the interface management system for orchestration according to the requirements of application development and front-end access.

[0016] Preferably, when the interface access system performs centralized management on the interfaces, the access frequency of each interface is counted every hour, and then each interface is divided into three levels: high, medium and low according to the access frequency. The classification criteria are as follows:

[0017] (1) The access frequency of the interface within the time range set on demand is Frei, and the frequency of each interface is normalized:

[0018]

[0019] (2) Interfaces are classified according to the value of normal_Fre:

[0020]

[0021] Preferably, when monitoring the interface, the interface access system counts each sub-interface parameter of the interface corresponding to the application request, the data returned by the background interface, and the time spent processing the request, and divides the sub-interfaces into three levels: high, medium, and low according to the time spent by each sub-interface in each time period.

[0022] Preferably, the interface access system establishes two recurrent neural networks during access optimization to predict the interface request and the specific content of the request, and caches the corresponding data in advance according to the prediction results, so as to improve the interface access speed, wherein:

[0023] The high-frequency access interface prediction model uses the RNN neural network to train the time series data generated by the interface level changes in the interface access management to obtain a model of the interface level changes in the system;

[0024] The high-frequency access interface request content prediction model predicts the high-frequency interface based on the high-frequency access interface prediction model. All historical access data of the high-level sub-interfaces of all sub-interfaces under the high-frequency interface are actually a multi-dimensional time-related sequence data.

[0025] Preferably, the recurrent neural network in the high-frequency access interface request content prediction model sets an RNN containing multiple serialized data inputs, an output layer, and 3 hidden layers, wherein each layer contains 256 nodes, and each input layer is connected to the nodes of the hidden layer, that is, the request parameters of the input sub-interface are output, and the request parameters of the interface at the next moment are output.

[0026] Preferably, the cloud computing API interface translation system is divided into a service layer, a management layer and an interface layer; the service layer provides a unified interface for applications, and forwards application requests to various background interfaces according to interface routing; the management layer provides management of the status of each interface, related information and interface orchestration; the interface layer links the interfaces of each platform.

[0027] The beneficial effects of the present invention are embodied in the following aspects:

[0028] (1) Centralized management interface facilitates management of the status information of each platform interface

[0029] Based on the abstract definition of the interface, standardized statistics of the interface information of the access system are realized, which facilitates the subsequent interface information management and the automatic arrangement of the interface.

[0030] (2) Providing a unified interface for corresponding applications through the orchestration interface

[0031] Use interface orchestration. All application requests are centrally processed through the interface access management system. It is not directly associated with the background interface, which facilitates the reuse of the background business interface and improves the flexibility of the entire application.

[0032] (3) Use recurrent neural networks to predict interface status and optimize interface requests

[0033] In the access management system, interfaces are divided into different levels according to the frequency of interface access and the processing time of different sub-interface requests. RNN neural network is used to learn high-level interface changes and sub-interface request parameters, thereby improving the speed of system processing application requests.

[0034] (4)Interfaces compatible with different transmission protocols

[0035] Use the interface access management system to convert different transmission protocols of different background interfaces into a unified transmission protocol, reducing the workload of application development for compatibility with different transmission protocols. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The overall structure of the system in the present invention is schematically shown;

[0037] Figure 2 The system flow chart of the present invention is schematically shown;

[0038] Figure 3 The RNN network interface diagram is schematically shown;

[0039] Figure 4 The network structure diagram of the RNN neural network is schematically shown. DETAILED DESCRIPTION

[0040] The following is a detailed description of embodiments of the present invention, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0041] The present invention provides a cloud computing API interface translation system, which belongs to the field of cloud computing and software engineering. The purpose is to uniformly manage interfaces, solve the problem that the number of API interfaces is huge and difficult to manage in the existing multi-platform development process, and arrange the interfaces according to the differences in functions implemented on different platforms, expose the interfaces on demand, convert interface data protocols, unify data protocols, and reduce the difficulty of application development.

[0042] The present invention arranges the interfaces required by the application to form a unified interface, and converts the interface data transmission protocol into the transmission protocol required by the application, realizes the separation of the background interface and the application, and uses the interface access system to uniformly manage and schedule the arranged interfaces. The entire system is divided into three core parts, including the interface management system, the interface arrangement system, and the interface access management system.

[0043] (1) Interface management system

[0044] The interface management system manages the interfaces of each platform in a unified manner, realizes the addition and deletion of interfaces, and manages the interface-related information. In the management system, each platform interface is defined as Interface (I1, I2, ... In; Type), In indicates the number of interfaces contained in the platform, and Type indicates the type of the platform; each sub-interface of the platform is defined as I (A (c), A (p), Desr, Tans, Tag), A (c) indicates the abstract list of front-end request services, A (p) indicates the abstract list of back-end response services, Desr indicates different interface descriptions, Tans indicates the transmission connection method, and Tag indicates the abstract identifier of the interface.

[0045] (2) Interface Orchestration System

[0046] The interface arrangement system automatically or manually arranges the interface according to the needs of application development and provides a unified interface to the application. The arrangement is to automatically combine the API interface or manually operate the interface to combine the API interface according to the needs of front-end translation words. The purpose and significance of the arrangement is to automatically and intelligently combine the provided API interfaces, improve the interface recognition ability and accuracy in the Internet environment, and better interpret the front-end needs and development applications.

[0047] In the automatic arrangement mode, according to a series of interfaces Interfacei (I1, I2, ... In) required for application development, and the services A (c) and transmission connection methods Tans required by each interface are defined, the similarity Sim (Ii, Ij) between each interface required by the application and the interfaces existing in the system is calculated in the interface management system (where Ii, Ij refer to two interfaces that need to be arranged), and a one-to-one correspondence is made according to the maximum similarity matching principle. The similarity calculation method is as follows:

[0048]

[0049] Parameter represents the interface parameter. The above formula represents the comparison and calculation of various parameters of the interface Ii required for application development and the interface Ij in the interface management system.

[0050]

[0051] in, Do not specify the parameters corresponding to the two interfaces that need to be orchestrated. Refers to the interface with the largest interface parameters among all the interfaces participating in the orchestration.

[0052] In manual orchestration mode, select the required interface in the interface management system for orchestration based on application development and front-end access requirements.

[0053] After the required interface is automatically or manually selected, a route for interface access is generated, and all selected interfaces are unified into an externally exposed interface according to the transmission protocol required by the application and managed by the interface access system.

[0054] (3) Interface access system

[0055] The interface access management system is used to centrally manage, monitor and optimize the access of the orchestrated interfaces. After processing the received request, it sends requests to various interfaces in the background, encapsulates the data returned by the background, and returns it to the application through a unified transmission protocol.

[0056] In terms of interface management, the system will count the access frequency of each interface every hour, and then divide each interface into three levels according to the access frequency: 1) high; 2) medium; 3) low. The classification criteria are as follows:

[0057] The access frequency of the interface within the time range set on demand is Frei, and the frequency of each interface is normalized:

[0058]

[0059] Interfaces are classified according to the normal_Fre value:

[0060]

[0061] In terms of interface monitoring, the system will count the parameters of each sub-interface of the interface corresponding to the application request, the data returned by the background interface, and the time spent processing the request. The sub-interfaces are divided into three levels according to the time spent on each sub-interface in each time period: 1) high; 2) medium; 2) low.

[0062] In the access optimization part, although the management system forwarding and processing time is less in a single interface access, the entire interface will still occupy more system resources in the case of high-frequency access. Since the access frequency and request processing data are a kind of time series data, two recurrent neural networks are established to predict the interface request and the specific content of the request. The system caches the corresponding data in advance according to the prediction results to improve the interface access speed.

[0063] a) High-frequency access interface prediction model

[0064] Time series data generated based on the interface level changes in interface access management. Use RNN neural network training to obtain a model of interface level changes in the system. The RNN network interface is as follows: Figure 3 Shown

[0065] b) High-frequency access interface request content prediction model

[0066] The high-frequency interface predicted based on the high-frequency access interface prediction model and all historical access data of the high-level sub-interfaces of all sub-interfaces under the interface are actually a multi-dimensional time-related sequence data.

[0067] Therefore, the present invention sets up an RNN including multiple serialized data inputs, an output layer, and 3 hidden layers, wherein each layer includes 256 nodes. Each input layer is connected to the nodes of the hidden layer. That is, the request parameters of the input sub-interface are output, and the request parameters of the interface at the next moment are output. The network structure diagram of the RNN neural network is as follows: Figure 4 shown.

[0068] The overall structure of the system in the present invention is divided into three levels: 1) Service layer: provides a unified interface for applications, and forwards application requests to various backend interfaces according to interface routing; 2) Management layer: provides management of each interface status, related information and interface arrangement; 3) Interface layer: links the interfaces of various platforms. Figure 1 shown.

[0069] The beneficial effects of the present invention are embodied in the following aspects:

[0070] (1) Centralized management interface facilitates management of the status information of each platform interface

[0071] Based on the abstract definition of the interface, standardized statistics of the interface information of the access system are realized, which facilitates the subsequent interface information management and the automatic arrangement of the interface.

[0072] (2) Providing a unified interface for corresponding applications through the orchestration interface

[0073] Use interface orchestration. All application requests are centrally processed through the interface access management system. It is not directly associated with the background interface, which facilitates the reuse of the background business interface and improves the flexibility of the entire application.

[0074] (3) Use recurrent neural networks to predict interface status and optimize interface requests

[0075] In the access management system, interfaces are divided into different levels according to the frequency of interface access and the processing time of different sub-interface requests. RNN neural network is used to learn high-level interface changes and sub-interface request parameters, thereby improving the speed of system processing application requests.

[0076] (4)Interfaces compatible with different transmission protocols

[0077] Use the interface access management system to convert different transmission protocols of different background interfaces into a unified transmission protocol, reducing the workload of application development for compatibility with different transmission protocols.

[0078] The innovations of the present invention are: abstract definition of the interface to facilitate the system's standardized statistics of interface information; design an interface orchestration system to provide different interface orchestration schemes for different applications; divide the platform interface into three levels according to the access frequency, and establish an RNN neural network model for prediction; use a recurrent neural network to predict the interface status and optimize the interface request.

[0079] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A cloud computing API interface translation system, characterized in that: include: Interface management system, interface orchestration system and interface access management system; The interface management system is used to uniformly manage the interfaces of each platform, realize the addition and deletion of interfaces, and manage the interface related information; include ， Define each platform interface as Interface(I1,I2,…I n ; Type), where I n Indicates the number of interfaces included in the platform, and Type indicates the type of the platform. Define each sub-interface of the platform as I(A(c), A(p), Desr, Tans, Tag), where A(c) represents the abstraction of the front-end request service list, A(p) represents the abstraction of the back-end response service list, Desr represents different interface descriptions, Tans represents the transmission connection mode, and Tag represents the abstract identifier of the interface; The interface arrangement system is used to automatically or manually arrange interfaces according to the needs of application development, generate interface access routes after automatically or manually selecting the required interfaces, and unify all selected interfaces into an externally exposed interface according to the transmission protocol required by the application and hand it over to the interface access management system for management; including: In the automatic arrangement mode, according to the series of interfaces required by the application development i (I1,I2,…I n ), and define the service A(c) and the transmission connection mode Tans required by each interface, and calculate the similarity sin(I) of each interface required by the application and the interface existing in the system in the interface management system. i ,I j ), one-to-one correspondence is performed according to the maximum similarity matching principle, and the similarity is calculated as follows: in, The above formula represents the interface I required for application development i Interface I in the interface management system j Parameter is compared and calculated, where Parameter represents the interface parameter, max(I_parameter num ) indicates the interface with the largest interface parameters among all the interfaces participating in the orchestration; or, In manual orchestration mode, select the required interface in the interface management system for orchestration according to the requirements of application development and front-end access; The interface access management system is used to centrally manage, monitor and optimize the access of the orchestrated interfaces. After processing the received requests, it sends requests to various interfaces in the background, encapsulates the data returned by the background, and returns it to the application through a unified transmission protocol; including: When the interface access management system centrally manages the interfaces, it counts the access frequency of each interface every hour, and then divides each interface into three levels: high, medium, and low according to the access frequency, including: the access frequency of the interface within the time range set on demand is Fre i , normalize the frequency of each interface and calculate the normal value according to the normalized value normal_Fre i Interfaces are classified into levels: In the formula, max(Fre) represents the maximum number of total interface accesses per hour, and min(Fre) represents the minimum number of total interface accesses per hour; The access optimization of the interface access management system establishes two recurrent neural networks to predict the interface request and the specific content of the request, and caches the corresponding data in advance according to the prediction results, so as to improve the interface access speed, wherein: The high-frequency access interface prediction model uses the RNN neural network to train the time series data generated by the interface level changes in the interface access management to obtain a model of the interface level changes in the system; The high-frequency access interface request content prediction model predicts the high-frequency interface based on the high-frequency access interface prediction model. All historical access data of the high-level sub-interfaces of all sub-interfaces under the high-frequency interface are multi-dimensional time-related sequence data.

2. The cloud computing API interface translation system according to claim 1, characterized in that: When monitoring the interface, the interface access management system counts each sub-interface parameter of the interface corresponding to the application request, the data returned by the background interface, and the time spent processing the request; and divides the sub-interfaces into three levels: high, medium, and low according to the time spent by each sub-interface in each time period.

3. The cloud computing API interface translation system according to claim 1, characterized in that: The recurrent neural network in the high-frequency access interface request content prediction model sets an RNN containing multiple serialized data inputs, an output layer, and 3 hidden layers, where each layer contains 256 nodes, and each input layer is connected to the nodes of the hidden layer, that is, the request parameters of the input sub-interface are output, and the request parameters of the interface at the next moment are output.

4. The cloud computing API interface translation system according to claim 3, characterized in that: The cloud computing API interface translation system is divided into a service layer, a management layer, and an interface layer; the service layer provides a unified interface for applications and forwards application requests to various backend interfaces according to interface routing; The management layer manages the status of each interface, related information, and interface orchestration; the interface layer links the interfaces of each platform.

Images