CN110780914B - Service publishing method and device - Google Patents

Abstract

Embodiments of the present invention provide a method and device for publishing a service. The method includes: obtaining a program file and a configuration file of the target service; generating a mirror file of the target service according to the program file and the configuration file; creating and starting an instance container of the target service according to the mirror file, The target service is provided externally through the instance container. In the embodiment of the present invention, the process of arrangement and release of services is simple and efficient, and can be applied to various types of services, providing unified service encapsulation, providing standardized service call interfaces externally, and realizing standardized encapsulation and integration of diversified models. Agile publishing; and based on containerized flexible services, the resources and capacity of instance containers can be dynamically adjusted to achieve flexible scheduling of target service resource usage and high maintainability.

Description

Service publishing method and device

technical field

The embodiments of the present invention relate to the technical field of mobile communication, and in particular, to a method and device for publishing a service.

Background technique

Artificial Intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. As a branch of computer science, AI aims to understand the essence of intelligence and produce an intelligent machine that can respond in a manner similar to human intelligence. Research in this field includes robotics, language recognition, image recognition, natural language processing and expert systems, etc. Since the birth of artificial intelligence, the theory and technology have become increasingly mature, and the application fields have also continued to expand. In the future, the technological products brought by artificial intelligence will be the "container" of human wisdom.

In the 1950s, the concept of AI first appeared, mainly based on expert systems; in the 1970s, with the invention of neural network algorithms, machine learning flourished, but limited by computing power, only shallow neural networks could be trained. With the advent of the era of cloud computing and big data, computing power and training data sets are no longer the bottleneck, and deep learning can achieve better accuracy than traditional machine learning algorithms with the increase of training data sets.

The deep integration of artificial intelligence technology represented by deep learning and the national economy has a very broad application prospect. In order to accelerate the application of the deep learning model, it is necessary to convert the deep learning model into a standard microservice and perform service arrangement and release; however, in the prior art, service arrangement and release are mainly implemented in the form of customized development and packaged into microservices When applied to specific business scenarios, this solution has the following problems:

1. Model release efficiency is low

Code development is required for each model release, and deep learning frameworks exist in various frameworks such as TensorFlow and Caffe, and developers have high learning costs.

2. Service standards are not uniform

The services developed by the existing technology are all personalized services, lack of unified standards, and increase the difficulty of application calls;

3. Inflexible service

Independently deployed personalized services lack an elastic response mechanism for irregular invocations of applications.

4. Poor service maintainability

For example, the invocation of services lacks a detailed statistical information system and a complete monitoring system.

Contents of the invention

Embodiments of the present invention provide a method and device for publishing services, which are used to solve the problems of low model publishing efficiency, non-uniform service standards and poor maintainability in service arrangement and publishing in the prior art.

On the one hand, an embodiment of the present invention provides a method for publishing a service, the method including:

Obtain the program file and configuration file of the target service;

Generate a mirror image file of the target service according to the program file and the configuration file;

According to the image file, an instance container of the target service is created and started, and the target service is provided externally through the instance container.

On the other hand, an embodiment of the present invention provides a service publishing device, the device comprising:

The acquisition module is used to acquire the program file and configuration file of the target service;

A file generating module, configured to generate a mirror file of the target service according to the program file and the configuration file;

The service issuing module is configured to create and start an instance container of the target service according to the image file, and provide the target service externally through the instance container.

On the other hand, an embodiment of the present invention also provides an electronic device, including a memory, a processor, a bus, and a computer program stored in the memory and operable on the processor. When the processor executes the program, the above-mentioned Steps in the service publishing method.

In yet another aspect, an embodiment of the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in the above service publishing method are implemented.

The service release method and device provided by the embodiments of the present invention obtain the program file and configuration file of the target service; generate the mirror file of the target service according to the program file and configuration file, and create and start the mirror file according to the mirror file The instance container of the target service creates an environment image in which the target service runs through the instance container, and provides the target service externally. In the embodiment of the present invention, the process of arrangement and release of services is simple and efficient, and can be applied to various types of services, providing unified service encapsulation, providing standardized service call interfaces externally, and realizing standardized encapsulation and integration of diversified models. Agile publishing; and based on containerized flexible services, the resources and capacity of instance containers can be dynamically adjusted to achieve flexible scheduling of target service resource usage and high maintainability.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

FIG. 1 is a schematic flowchart of a service publishing method provided by an embodiment of the present invention;

FIG. 2 is a block diagram of the service release architecture of the first example of the embodiment of the present invention;

FIG. 3 is a block diagram of a service publishing device of a second example of an embodiment of the present invention;

Fig. 4 is the method flowchart of the second example of the embodiment of the present invention;

FIG. 5 is a block diagram of a service publishing device provided by an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.

detailed description

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Fig. 1 shows a schematic flowchart of a method for publishing a service provided by an embodiment of the present invention.

As shown in Figure 1, the service release method provided by the embodiment of the present invention, the method specifically includes the following steps:

Step 101, acquire the program file and configuration file of the target service.

Among them, the target service is the service to be arranged and/or published in the current operating environment, which can be an AI service matured after deep learning training, or other atomic services; in the embodiment of the present invention, the atomic service refers to the smallest service in the current operating environment service unit. The target service has different corresponding program files according to its attributes. For example, for model services, the program files are service model files, including models such as calculation graphs and data flows; for application services, the program files are application program codes Bag.

The configuration file includes the parameters of the target service, such as the usage framework of the target service, the list of input/output parameters conforming to the preset interface specification, and so on.

As a first example, refer to FIG. 2. FIG. 2 provides a service publishing architecture (hereinafter referred to as architecture), which provides service orchestration and publishing functions, and obtains program files and configurations of target services during the process of orchestrating services. document.

Step 102, generate a mirror image file of the target service according to the program file and the configuration file.

Wherein, the image file is a file generated by a program file and a configuration file, and is used to provide the target service. The program files and configuration files are pre-processed and packaged into an image file of the target service, so as to create an environment image in which the target service runs.

Step 103: Create and start an instance container of the target service according to the image file, and provide the target service externally through the instance container.

Among them, the instance container is a virtual container used to run the instance. As a resource sharing method, the instance container can provide great flexibility. According to the image file, the instance container is created and started. During the process of creating the instance container, the program file is processed into a service that can be called externally and internally, and the processed service is merged into the previous operating environment image. Finally, It is encapsulated into a basic image containing the current operating environment of the target service, and then in the current operating environment, the target service can be provided externally through the instance container, so that when the user needs to run the target service later, he only needs to start the instance container. Yes, there is no need to carry out underlying development work such as code development, and improve service arrangement and release timeliness.

And based on containerized instance deployment, it can realize flexible services, provide elastic expansion and contraction functions, and monitor the response of target services and the resource overhead of each instance container, and finally realize the dynamic adjustment of the number of containers. Optionally, the instance container can be a Docker container or a Mesos container.

Continuing to refer to FIG. 2 , there are three instance containers in FIG. 2 , which are containers for model class encapsulation, application class encapsulation, and service class encapsulation.

In the above embodiments of the present invention, by obtaining the program file and configuration file of the target service; according to the program file and configuration file, the image file of the target service is generated, and the target service is created and started according to the image file The instance container of the target service is created through the instance container to create an environment image for running the target service, and the target service is provided externally. In the embodiment of the present invention, the process of arrangement and release of services is simple and efficient, and can be applied to various types of services, providing unified service encapsulation, providing standardized service call interfaces externally, and realizing standardized encapsulation and integration of diversified models. Agile publishing; and based on containerized flexible services, the resources and capacity of instance containers can be dynamically adjusted to achieve flexible scheduling of target service resource usage and high maintainability. The embodiment of the present invention solves the problems of low model release efficiency, non-uniform service standards, poor maintainability and the like existing in service arrangement and release in the prior art.

Optionally, in the embodiment of the present invention, the step of acquiring the program file and configuration file of the target service includes:

When the attribute of the target service is model class encapsulation or application class encapsulation, the program file and the configuration file of the target service are received.

Wherein, when the attribute (type) of the target service is model class encapsulation or application class encapsulation, the program file and configuration file of the target service are received, wherein the received program file and configuration file may be uploaded by the user, or may be uploaded by a third-party device.

Specifically, the target service of the model class encapsulation is the encapsulation based on the algorithm model; the application class encapsulation is used to combine logic algorithms between multiple atomic service models, preprocessing logic algorithms before model algorithms, and other scenarios.

Referring to Fig. 2, each instance container in Fig. 2 corresponds to target services of different attributes, including model class encapsulation, application class encapsulation and service class encapsulation. Optionally, the architecture shown in FIG. 2 may provide a user interface (UserInterface, UI) for the user to select the attribute of the target service when uploading the program file and/or configuration file.

Optionally, in the embodiment of the present invention, after the step of receiving the program file and configuration file of the target service, the method further includes:

According to the configuration file, the program file is converted into a first format file conforming to the remote procedure call RPC interface protocol and a second format file conforming to the hypertext transfer protocol HTTP interface protocol.

Among them, in order to provide a standardized service call interface externally, after receiving the program file and configuration file of the target service, the program file is processed to conform to the Remote Procedure Calls (RPC) according to the input/output parameters in the configuration file. The first format file of the interface protocol is for remote calling; and the second format file conforming to the Hyper Text Transfer Protocol (Hyper Text Transfer Protocol, HTTP) interface protocol provides standard HTTP interface services and internal application calls.

It is understandable that when the attribute of the target service is encapsulated by the application class, if the algorithm itself in the program file has realized the output of the HTTP interface, it is not necessary to convert it to the second format file at this time, and only needs to convert its interface to the HTTP interface. It is enough to standardize the input/output format.

Provide a unified service package, provide a standardized service call interface to the outside world, and realize the standardized package and agile release of diversified models.

Referring to Fig. 2, after the program files and configuration files are uploaded through the standard Representational State Transfer (REST) interface, the framework internally converts the program files into files in RPC interface or HTTP interface format.

Optionally, in the embodiment of the present invention, when the attribute of the target service is encapsulated by the model class, the program file is a model file of the target service, and the model file includes: a calculation graph, a data flow, Variable parameters and/or signatures; the model usage framework of the model file, the interface specification and/or the configuration parameters of the target service;

When the attribute of the target service is encapsulated by the application class, the program file is a program code file of the target service; the configuration file includes: a code language format and/or a dependency package of the program code file.

On the one hand, for model class encapsulation, the program file is a model file serving the target, and the model file includes: calculation graph, data flow, variable parameters and/or signature of the algorithm model, and may also include some auxiliary files, The model usage framework of the model file, the interface specification and/or the configuration parameters of the target service, wherein the interface specification includes the input/output parameter list data flow conforming to the interface specification; optionally, the configuration file may also include Whether the target service requires special processing flags and the configuration of operating parameters related to publishing the model as a microservice, etc.

On the other hand, for application class encapsulation, the program file is the program code file of the target service, and the configuration file includes the code language format of the program code file and/or parameters such as dependent packages, which are used to create the adaptation target service The image of the environment where the logic algorithm runs.

Optionally, in the embodiment of the present invention, the step of acquiring the program file and configuration file of the target service includes:

When an orchestration request for service orchestration is received, obtain the atomic service to be orchestrated and the configuration information of the atomic service;

Generate a calculation graph configuration file according to the atomic service and the configuration information;

Obtaining the preset program files of the atomic services, combining the program files of the atomic services to obtain the program files of the target service, and combining the configuration files according to the configuration information to obtain the configuration files of the target services.

As another implementation manner, when an orchestration request for service orchestration is received, the orchestration request may be a user's deployment of an atomic service icon.

Obtain the atomic service to be arranged in the orchestration request and the configuration information of the atomic service, and convert it into a calculation graph configuration file of the atomic service; the calculation graph configuration file includes the list of atomic services required by the total service (target service), and the configuration of each atomic service Information, the configuration information includes the input/output parameter list of the atomic service, and the corresponding relationship between the input/output parameters of each atomic service.

After the calculation graph configuration file is generated, obtain the preset program files of each atomic service, and combine the program files of the target service; and combine the configuration files of the target service according to the configuration information, and combine the atomic services Got target service. The service orchestration process is simple and efficient, and can be applied to various types of services.

Optionally, in the embodiment of the present invention, the step of creating and starting the instance container of the target service includes:

Create and start an instance container of the target service according to the configuration number of the instance container included in the configuration file;

or

Create and start a preset number of instance containers of the target service.

Among them, container-based instance deployment can realize flexible services and provide elastic expansion and contraction functions. The configuration number of instance containers can be determined according to the limit in the configuration file, and can also be preset to a preset number of instance containers; and It can monitor the response of the target service and the resource overhead of each instance container to dynamically adjust the number of containers.

Optionally, in the embodiment of the present invention, after the step of creating and starting the instance container of the target service, the method further includes:

According to the configuration file, allocate graphics processor GPU video memory resources to the instance container, and monitor the GPU state information of the instance container.

Among them, according to the configuration file, the graphics processor (Graphics Processing Unit, GPU) video memory resource can be allocated to the instance container. The allocation of GPU resources supports fine-grained segmentation based on video memory, and the GPU status information of the instance container can be monitored. .

Specifically, the monitoring of GPU status information can adopt the Master-slave architecture. For example, deploy slave services on each instance container equipped with GPUs to monitor GPU memory resources on the instance container, and each slave service sends real-time monitoring information to the Master. Make a summary. The Master controls the GPU resources of the entire device and responds to the GPU application request from the application side to allocate and schedule the GPU. After the application is running, the Master updates and recycles the GPU resources in time through the monitoring information reported by the Slave to complete the allocation of the GPU resources of the entire architecture. Control and scheduling.

Optionally, appropriate CPU resources can also be allocated to the instance container according to the characteristics of the service to achieve ideal performance.

In the above embodiments of the present invention, by obtaining the program file and configuration file of the target service; according to the program file and configuration file, the image file of the target service is generated, and the image file of the target service is created and started according to the image file The instance container creates an environment image for running the target service through the instance container, and provides the target service externally. In the embodiment of the present invention, the process of arrangement and release of services is simple and efficient, and can be applied to various types of services, providing unified service encapsulation, providing standardized service call interfaces externally, and realizing standardized encapsulation and integration of diversified models. Agile publishing; and based on containerized flexible services, the resources and capacity of instance containers can be dynamically adjusted to achieve flexible scheduling of target service resource usage and high maintainability.

As a second example, the service release method provided by the present invention will be introduced below with reference to FIG. 3 and FIG. 4 .

Wherein, a service publishing device 300 is provided in FIG. 3, including: an encapsulation engine 301, an arrangement engine 302, and a load balancing module 303; wherein, the encapsulation engine 301 includes an encapsulation processing module 3011 and a microservice conversion module 3012; the arrangement engine 302 includes Visualization processing module 3021.

Referring to FIG. 4, after the service publishing process is started, the working process of the service publishing device 300 mainly includes the following steps:

Step 401, determine whether the attribute of the target service is service orchestration: if not, execute step 402;

If so, execute step 403, obtain the atomic service to be arranged and the configuration information of the atomic service, generate a calculation graph configuration file, and execute step 406;

Among them, the visualization processing module 3021 receives the deployment of the atomic service icon and converts it into a calculation graph configuration file of the atomic service, including configuration information defined for the total service, a list of required atomic services, a list of input/output parameters of each atomic service, and a list of each atomic service Correspondence between input/output parameters between services, and transfer the calculation graph configuration file to the encapsulation processing module 3011. The encapsulation processing module 3011 will create an environment mirror adapted to the operation of advanced algorithms according to the operation dependent parameters of each atomic service in the input calculation graph configuration file, and transfer the calculation graph configuration file to the microservice conversion module 3012.

Step 402, judging whether the attribute of the target service is service class encapsulation: if not, go to step 405;

If yes, execute step 404, receive the configuration file, and execute step 406;

Step 405, determine that the attribute of the target service is model class encapsulation or application class encapsulation, receive program files and configuration files, and execute step 406;

Wherein, the encapsulation processing module 3011 receives program files and configuration files.

Step 406, converting the program file into a first format file conforming to the RPC interface protocol and a second format file conforming to the HTTP interface protocol, generating an image file of the target service, and performing step 407;

Among them, the microservice conversion module 3012 converts the program file into the first format file and the second format file and sends it back to the encapsulation processing module 3011, which is used for the encapsulation processing module 3011 to generate the image file of the target service and create the basis of the operating environment mirror image.

Step 407, creating and starting an instance container of the target service.

Step 408, register the instance container with the load balancing module 303, and provide the target service externally.

Further, the following uses AI service as an example and target services with different attributes to introduce the working process of the service publishing apparatus 300 in FIG. 3 .

(1) Model class encapsulation

The model class encapsulation executes step 401, step 402, step 405, step 406, step 407 and step 408 in Fig. 4 in sequence;

Specifically, model class encapsulation is mainly aimed at the encapsulation of core model algorithms in AI general service capabilities. Model developers upload model files and configuration files to the encapsulation processing module 3011; wherein, model files include calculation graphs, data flows, and input and output of related variables, signatures, and auxiliary files, etc., and configuration files include model usage frameworks, conforming to device interface specifications, etc. The list of input/output parameters, whether the data stream needs special processing flags, and the configuration of relevant operating parameters for the model to be published as microservices, etc.; the encapsulation processing module 3011 creates an environment mirror for the operation of the adaptive model algorithm according to the parameters such as the model usage framework.

In the next step, configuration files and model files will be passed to the microservice conversion module 3012, which will process the model files into RPC interface and standard HTTP interface services available for remote calls according to the input/output parameters in the configuration, and return the encapsulated Processing module 3011. The encapsulation processing module 3011 merges the processed service into the previous operating environment image, and finally encapsulates it into a basic image containing the service and the operating environment.

Thereafter, the basic image and configuration file will be input into the load balancing module 303, and according to the service instance operating parameter N input in the model configuration (if not set, it is the default value of the device), the load balancing module 303 will create N running service instances based on the image Containers are automatically registered with the load balancer.

Finally, the load balancing module 303 will output a unified interface that can provide the AI microservice, so that it can be conveniently called by AI developers in different scenarios for rapid development. At the same time, the load balancing module 303 will dynamically expand and shrink the number of containers running service instances according to the call request load, so as to achieve efficient dynamic allocation of resources.

(2) Application class encapsulation

The application class encapsulation executes step 401, step 402, step 405, step 406, step 407 and step 408 in Fig. 4 in sequence;

Application class encapsulation is mainly aimed at the encapsulation of key logic algorithms in AI service capabilities other than core model algorithms, and is usually used in scenarios such as combining logic algorithms between multiple atomic models and preprocessing logic algorithms before model algorithms.

The model developer uploads the corresponding program code files and configuration files to the encapsulation processing module 3011, which creates an environment image for the operation of the adaptive logic algorithm according to the code language format, dependent packages and other parameters.

Then, the program code file and configuration file will be transferred to the microservice conversion module 3012, and the subsequent process (step 406, step 407, step 408) is the same as the above-mentioned model class encapsulation, and will not be described in this embodiment of the present invention.

(3) Service class encapsulation

The service class encapsulation executes step 401, step 402, step 404, step 406, step 407 and step 408 in Fig. 4 in turn;

The service class is mainly aimed at introducing mature AI services from the outside and standardizing unified integration through device interfaces. It is only necessary to upload the configuration file describing external service-related parameters (such as access address, input/output parameter list, whether parameters need to be converted flags, and publishing microservice-related operating parameter configurations, etc.) through the service encapsulation engine 301 of the device, and the encapsulation process Module 3011 will encapsulate the AI service into a basic image that contains the service's operating environment.

Then, the basic image and the configuration file will be input into the load balancing module 303, and the subsequent process (step 406, step 407, step 408) is the same as the above-mentioned model class encapsulation, and will not be repeated here in the embodiment of the present invention.

(4) Service orchestration

Model class encapsulation executes step 401, step 403, step 406, step 407, step 408 in the above-mentioned figure 4 in sequence;

Through the encapsulation processing module 3011, the module algorithms of the target services of the above attributes will be encapsulated into a series of AI atomic services. For advanced services that need to combine several atomic services, the orchestration engine 302 of this device will provide a visual service orchestration function.

The orchestration engine 302 receives the configuration information of the AI atomic service, and converts it into a computing graph configuration file of the atomic service. The computing graph configuration file includes a list of atomic services required by the total service (target service), configuration information of each atomic service, and the configuration information includes the atomic service. The input/output parameter list of the service, and the corresponding relationship between the input/output parameters of each atomic service.

The encapsulation processing module 3011 will create an environment mirror adapted to the operation of advanced algorithms according to the operation dependent parameters of each atomic service in the calculation graph configuration file, and pass the calculation graph configuration file to the microservice conversion module 3012.

The microservice conversion module 3012 organically associates the relevant atomic services according to the correspondence between the atomic services in the calculation graph configuration file, and transfers them to the encapsulation processing module 3011, which is encapsulated into a general AI service interface with unified input/output and A high-level service image of its operating environment.

Then, the high-level service image and the calculation graph configuration file will be input into the load balancing module 303, and the subsequent process (step 406, step 407, step 408) is the same as the above-mentioned model class encapsulation, and the embodiment of the present invention will not repeat them here.

In addition, the device shown in Figure 3 also provides the following functions:

(1) Multi-version switching: AI service is a forward-looking research, which requires continuous trial and error iterations to achieve the desired effect. In order to realize seamless switching between multiple service versions, this device provides a grayscale release function, which can realize smooth online upgrade of services without completely terminating the service.

(2) Service-level monitoring: This device provides platform-level, service-level and container-level fine-grained monitoring functions. Service failures can be traced back to specific hosts and containers.

(3) Multi-user management: This device supports multi-user management function. Users can apply for hardware resources such as CPU and GPU, and create and manage their own applications and services.

In the above example, the online release platform, underlying implementation, and resource configuration are independent, and the online agile release of various deep learning framework models in different scenarios is realized. The release efficiency is fast, so that modelers can focus on model training without paying attention to the model It can accelerate the efficiency of applying AI models to production; support the encapsulation of model classes, application classes, and service classes, and seamlessly switch between multi-model versions without business awareness. For the realization of different versions of AI services, they can be provided through the online service platform The gray-scale publishing capability realizes service upgrades that are not perceived by the business; the service can be monitored in real time and faults can be traced back, and the operation and maintenance personnel can perceive the abnormal state of the service at the first time, and call the complete service provided by the online service platform Logs can quickly locate specific service instances and have high maintainability; through the hybrid scheduling of CPU and GPU, dynamic expansion and contraction, etc., the scalability is strong, and it can provide high-availability, high-concurrency, and low-latency online for production. Serve.

The service publishing method provided by the embodiment of the present invention is described above, and the service publishing device provided by the embodiment of the present invention will be described below with reference to the accompanying drawings.

Referring to Fig. 5, an embodiment of the present invention provides a device for publishing a service, which includes:

The acquisition module 501 is configured to acquire program files and configuration files of the target service.

Among them, the target service is the service to be arranged and/or published in the current operating environment, which can be an AI service matured after deep learning training, or other atomic services; in the embodiment of the present invention, the atomic service refers to the smallest service in the current operating environment service unit. The target service has different corresponding program files according to its attributes. For example, for model services, the program files are service model files, including models such as calculation graphs and data flows; for application services, the program files are application program codes Bag.

The configuration file includes the parameters of the target service, such as the usage framework of the target service, the list of input/output parameters conforming to the preset interface specification, and so on.

The file generation module 502 is configured to generate a mirror file of the target service according to the program file and the configuration file.

Wherein, the image file is a file generated by a program file and a configuration file, and is used to provide the target service. The program files and configuration files are pre-processed and packaged into an image file of the target service, so as to create an environment image in which the target service runs.

The service issuing module 503 is configured to create and start an instance container of the target service according to the image file, and provide the target service externally through the instance container.

Among them, the instance container is a virtual container used to run the instance. As a resource sharing method, the instance container can provide great flexibility. According to the image file, the instance container is created and started. During the process of creating the instance container, the program file is processed into a service that can be called externally and internally, and the processed service is merged into the previous operating environment image. Finally, It is encapsulated into a basic image containing the current operating environment of the target service, and then in the current operating environment, the target service can be provided externally through the instance container, so that when the user needs to run the target service later, he only needs to start the instance container. Yes, there is no need to carry out underlying development work such as code development, and improve service arrangement and release timeliness.

And based on containerized instance deployment, it can realize flexible services, provide elastic expansion and contraction functions, and monitor the response of target services and the resource overhead of each instance container, and finally realize the dynamic adjustment of the number of containers. Optionally, the instance container can be a Docker container or a Mesos container.

Optionally, in the embodiment of the present invention, the acquiring module 501 includes:

The receiving sub-module is used to receive the program file and configuration file of the target service when the attribute of the target service is packaged by model class or packaged by application class.

Optionally, in the embodiment of the present invention, the device further includes:

The format conversion module is used to convert the program file into a first format file conforming to the remote procedure call RPC interface protocol and a second format file conforming to the hypertext transfer protocol HTTP interface protocol according to the configuration file.

Optionally, in the embodiment of the present invention, when the attribute of the target service is encapsulated by the model class, the program file is a model file of the target service, and the model file includes: a calculation graph, a data flow, Variable parameters and/or signatures; the model usage framework of the model file, the interface specification and/or the configuration parameters of the target service;

When the attribute of the target service is encapsulated by the application class, the program file is a program code file of the target service; the configuration file includes: a code language format and/or a dependency package of the program code file.

Optionally, in the embodiment of the present invention, the acquiring module 501 includes:

The obtaining submodule is used to obtain the atomic service to be arranged and the configuration information of the atomic service when receiving an arrangement request for service arrangement;

Generate a calculation graph configuration file according to the atomic service and the configuration information;

Obtaining the preset program files of the atomic services, combining the program files of the atomic services to obtain the program files of the target service, and combining the configuration files according to the configuration information to obtain the configuration files of the target services.

Optionally, in the embodiment of the present invention, the service publishing module 503 includes:

The first creating submodule is used to create and start the instance container of the target service according to the configuration number of the instance container included in the configuration file;

or

The second creating submodule is used to create and start a preset number of instance containers of the target service.

Optionally, in the embodiment of the present invention, the device further includes:

The resource allocation module is configured to allocate graphics processor GPU memory resources to the instance container according to the configuration file, and monitor GPU state information of the instance container.

In the above-mentioned embodiments of the present invention, the program file and configuration file of the target service are obtained through the acquisition module 501; the file generation module 502 generates the image file of the target service according to the program file and configuration file, and the service publishing module 503 generates the image file of the target service according to the An image file, creating and starting an instance container of the target service, creating an environment image for running the target service through the instance container, and providing the target service externally. In the embodiment of the present invention, the process of arrangement and release of services is simple and efficient, and can be applied to various types of services, providing unified service encapsulation, providing standardized service call interfaces externally, and realizing standardized encapsulation and integration of diversified models. Agile publishing; and based on containerized flexible services, the resources and capacity of instance containers can be dynamically adjusted to achieve flexible scheduling of target service resource usage and high maintainability.

Fig. 6 shows a schematic structural diagram of an electronic device provided by another embodiment of the present invention.

Referring to FIG. 6 , an electronic device provided by an embodiment of the present invention includes a memory (memory) 61, a processor (processor) 62, a bus 63, and a computer program stored in the memory 61 and operable on the processor. Wherein, the memory 61 and the processor 62 communicate with each other through the bus 63 .

The processor 62 is configured to call the program instructions in the memory 61, so as to implement the methods provided in the above-mentioned embodiments of the present invention when executing the program.

In another implementation manner, the processor implements the following method when executing the program:

Obtain the program file and configuration file of the target service;

Generate a mirror image file of the target service according to the program file and the configuration file;

According to the image file, an instance container of the target service is created and started, and the target service is provided externally through the instance container.

The electronic device provided by the embodiment of the present invention can be used to execute the program corresponding to the method in the foregoing method embodiment, and details will not be described in this embodiment.

The electronic device provided by the embodiment of the present invention obtains the program file and configuration file of the target service; generates the image file of the target service according to the program file and the configuration file, and creates and starts the target service according to the image file The instance container of the service creates an environment image for running the target service through the instance container, and provides the target service externally. In the embodiment of the present invention, the process of arrangement and release of services is simple and efficient, and can be applied to various types of services, providing unified service encapsulation, providing standardized service call interfaces externally, and realizing standardized encapsulation and integration of diversified models. Agile publishing; and based on containerized flexible services, the resources and capacity of instance containers can be dynamically adjusted to achieve flexible scheduling of target service resource usage and high maintainability.

Another embodiment of the present invention provides a non-transitory computer-readable storage medium, in which a computer program is stored on the non-transitory computer-readable storage medium, and when the program is executed by a processor, the above-mentioned embodiments of the present invention are implemented The steps in the method provided in .

In another implementation manner, when the program is executed by the processor, the following methods are implemented:

Obtain the program file and configuration file of the target service;

Generate a mirror image file of the target service according to the program file and the configuration file;

According to the image file, an instance container of the target service is created and started, and the target service is provided externally through the instance container.

In the non-transitory computer-readable storage medium provided by the embodiments of the present invention, when the program is executed by a processor, the method in the foregoing method embodiment is implemented, and details will not be repeated in this embodiment.

The non-transitory computer-readable storage medium provided by the embodiment of the present invention obtains the program file and configuration file of the target service; generates the image file of the target service according to the program file and configuration file, and according to the image file, An instance container of the target service is created and started, an environment image for running the target service is created through the instance container, and the target service is provided externally. In the embodiment of the present invention, the process of arrangement and release of services is simple and efficient, and can be applied to various types of services, providing unified service encapsulation, providing standardized service call interfaces externally, and realizing standardized encapsulation and integration of diversified models. Agile publishing; and based on containerized flexible services, the resources and capacity of instance containers can be dynamically adjusted to achieve flexible scheduling of target service resource usage and high maintainability.

Yet another embodiment of the present invention discloses a computer program product, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by a computer When, the computer can execute the method provided by the above method embodiments, for example including:

When the failed switchback request sent by the terminal is received, the target cell corresponding to the failed switchback request is determined; the failed switchback request is a reset request after the terminal performs enhanced single wireless voice call continuity ESRVCC switching execution failure;

determining the number of failed handover requests corresponding to the target cell;

When the number of requests reaches the preset threshold of the target cell, stop forwarding the ESRVCC handover request for the target cell.

The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. It can be understood and implemented by those skilled in the art without any creative effort.

Through the above description of the implementations, those skilled in the art can clearly understand that each implementation can be implemented by means of software plus a necessary general-purpose hardware platform, and of course also by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of software products, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic discs, optical discs, etc., including several instructions to make a computer device (which may be a personal computer, server, or network device, etc.) execute the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims (7)

1. A service release method, characterized in that, comprising: Obtain the program file and configuration file of the target service; Generate a mirror image file of the target service according to the program file and the configuration file; Create and start an instance container of the target service according to the image file, and provide the target service externally through the instance container; The steps of obtaining the program file and the configuration file of the target service include: When receiving an orchestration request for service orchestration, acquiring the atomic service to be orchestrated in the orchestration request and the configuration information of the atomic service; Generate a calculation graph configuration file according to the atomic service and the configuration information, and the calculation graph configuration file includes a list of atomic services required by the target service and configuration information of each atomic service; Based on the calculation graph configuration file, obtain the preset program file of the atomic service, combine the program file of the atomic service to obtain the program file of the target service, and combine the program file of the target service according to the configuration information to obtain the target service configuration file; The steps of obtaining the program file and the configuration file of the target service include: When the attribute of the target service is model class encapsulation or application class encapsulation, receiving the program file and configuration file of the target service; When the attribute of the target service is encapsulated by the model class, the program file is a model file of the target service, and the model file includes: calculation graph, data flow, variable parameters and/or signature; the model Documented model usage framework, interface specification and/or configuration parameters of said target service; When the attribute of the target service is encapsulated by the application class, the program file is a program code file of the target service; the configuration file includes: a code language format and/or a dependency package of the program code file. 2. The method according to claim 1, characterized in that, after the step of receiving the program file and the configuration file of the target service, the method further comprises: According to the configuration file, the program file is converted into a first format file conforming to the remote procedure call RPC interface protocol and a second format file conforming to the hypertext transfer protocol HTTP interface protocol. 3. The method according to claim 1, wherein the step of creating and starting the instance container of the target service comprises: Create and start an instance container of the target service according to the configuration number of the instance container included in the configuration file; or Create and start a preset number of instance containers of the target service. 4. The method according to claim 1, wherein after the step of creating and starting the instance container of the target service, the method further comprises: According to the configuration file, allocate graphics processor GPU video memory resources to the instance container, and monitor the GPU state information of the instance container. 5. A service release device, characterized in that it comprises: The acquisition module is used to acquire the program file and configuration file of the target service; A file generating module, configured to generate a mirror file of the target service according to the program file and the configuration file; A service publishing module, configured to create and start an instance container of the target service according to the image file, and provide the target service externally through the instance container; The acquisition module includes an acquisition submodule, and the acquisition submodule is used for: When receiving an orchestration request for service orchestration, obtain the atomic service to be orchestrated in the orchestration request and the configuration information of the atomic service; Generate a calculation graph configuration file according to the atomic service and the configuration information, and the calculation graph configuration file includes a list of atomic services required by the target service and configuration information of each atomic service; Based on the calculation graph configuration file, obtain the preset program file of the atomic service, combine the program file of the atomic service to obtain the program file of the target service, and combine the program file of the target service according to the configuration information to obtain the target service configuration file; The acquisition module includes: The receiving submodule is used to receive the program file and configuration file of the target service when the attribute of the target service is model class encapsulation or application class encapsulation; When the attribute of the target service is encapsulated by the model class, the program file is a model file of the target service, and the model file includes: calculation graph, data flow, variable parameters and/or signature; the model Documented model usage framework, interface specification and/or configuration parameters of said target service; When the attribute of the target service is encapsulated by the application class, the program file is a program code file of the target service; the configuration file includes: a code language format and/or a dependency package of the program code file. 6. An electronic device, characterized in that it comprises a memory, a processor, a bus, and a computer program stored in the memory and operable on the processor, when the processor executes the program, it realizes the requirements of claims 1 to 4. The steps in any one of the service publishing methods. 7. A non-transitory computer-readable storage medium, on which a computer program is stored, characterized in that: when the program is executed by a processor, the service release method according to any one of claims 1 to 4 is implemented A step of.

Images