Due to the wide spread of cloud computing, arises actual question about architecture, design and implementation of cloud applications. The microservice model describes the design and development of loosely coupled cloud applications when computing resources are provided on the basis of automated IaaS and PaaS cloud platforms. Such applications consist of hundreds and thousands of service instances, so automated validation and testing of cloud applications developed on the basis of microservice model is a pressing issue. There are constantly developing new methods of testing both individual microservices and cloud applications at a whole. This article presents our vision of a framework for the validation of the microservice cloud applications, providing an integrated approach for the implementation of various testing methods of such applications, from basic unit tests to continuous stability testing.
The microservice model describes a cloud application as a suite of small
independent services, each running in its own container and communicating with other
services using lightweight mechanisms. These services are built around separate
business capabilities, independently deployable and may be written by di erent
development teams using di erent programming languages and frameworks [
One of the most important stages of the process of continuous integration is a continuous validation of the software. The goal of this study is to develop a microservice validation methodology and design a software solution for automated microservice systems testing support. To achieve this goal it is necessary accomplish the following tasks: { provide a review of existing distributed systems validation methods, including cloud, multi-agent and actor-based systems; { develop a microservice validation methodology; { design a framework for microservice systems validation support. In this article, we will describe our model of validation of microservice systems and it's implementation as a prototype of microservice systems validation framework.
The rest of this paper is organized as follows. In Section 2 we will provide a review of the current state of distributed systems testing and validation methods. In Section 3 we would describe the model of validation of microservice systems. In Section 4 we would propose an architecture and describe prospective approaches to implementation of microservice systems validation framework. In Section 5 we would summarize the results of our research and discuss further research directions. 2
To develop a model of microservice systems validation, we should analyze
existing testing methods that used for such distributed systems. In the paper [
We also analyze the validation methods that are used in a set of related
distributed systems models, like multi-agent systems and actors. For example, it is
possible to apply ontologies-based multi-agent systems validation method [
We should also mention integration and stability testing methods are
developed for highly loaded cloud applications. The \Chaos Monkey" [
The analysis of existing methods of testing of microservice systems shows, that there is no uni ed model of microservice system validation that would support a continuous integration of microservices.
Based on the microservice features shown in the introduction, we propose the following model of microservice systems validation (see g. 1). 1. To provide an automated validation of microservices interface and communication process, we should require a de nition of the interface of every microservice. 2. According with the features of the programming language and framework chosen to implement the microservice, a developer should provide appropriate automated unit-testing of the microservice source code to ensure it's compliance with the requirements. 3. If the source code is passing all the unit tests, the microservice is packed into the container and a set of container self-tests is provided to ensure that all the components of the container are functioning correctly and the interface of the container corresponds to the interface de nition provided on the step 1. 4. If the self-test is successful, then microservice load testing and security testing is provided. Steps 1-4 can be provided locally on a microservice developer's machine. 5. If all tests pass, the microservice is deployed on a test microservice environment, where functional integration testing, load integration testing and security integration testing is performed. It allow to detect a set of issues caused by microservices orchestration, including load balancing, life cycle management and communication issues. 6. If the microservice is passing all tests inside the test environment, it can be deployed to the production environment. A continuous stability testing is performed in a production environment in accordance with the methodology of deliberate provocation of random failures of the system components. 4
To evaluate the proposed validation model, a Microservice Systems Validation Framework (MSVF) is being developed. The MSVF should support microservice application testing from the source code to the continuous stability testing, regardless of the basic programming languages and software frameworks used for the microservice development (see g. 2).
MSVF would provide its users a catalog of testing methods, that can be used to test a microservice application and an API for integration with continuous integration platforms. Each testing method can be considered as a separate microservice template, that can be tuned to provide validation of speci c part or activity of user's microservice application.
We can de ne the following main actors, who would interact with the MSVF: { Validation methods developer: provides development of new validation methods and their integration into the MSVF. The validation method can be implemented as an independent microservice, that implements a speci c method of validation that can be adopted to a speci c microservice or microservice application, including: component-level testing: supports unit-testing of the microservice source code and container self-testing; integration-level testing: supports functional integration testing of the microservice system on a testing environment; load testing : supports load integration testing of the microservice system on a testing environment; stability testing : supports validation methods that provides deliberate provocation of random failures of the microservice system in a production environment. { Tester: is the main user of the MSVF platform. Tester can choose a service or microservice application for validation; de ne a validation objective that consists of a set of validation methods, adapted to the chosen application; execute validation objectives and gather the results of the validation. { Microservice application: represents the validated microservice application.
We can de ne the following basic components of the MSVF (see g. 3): { Validation Methods Manager responds for listing, creation and modi cation of validation methods available in the MSVF. { Validation Method Interface is a common API that should be provided by any validation method in the MSVF to support adaptation of validation method to a speci c microservice application. { Validation Method is an implementation of speci c validation method in the form of microservice template. { Validation Objectives Manager responds for listing, creation and modi cation of validation objectives. { Validation Objectives Interface de nes an interface of validation objectives, including methods for integration and tuning of validation methods. { Validation Objective represents a suite of validation methods, tuned for a speci c microservice application.
Currently, the MSVF is being developed to support VAMP and Mjolnirr microservice platforms and Jenkins as a continuous integration framework. VAMP and Mjolnirr cloud infrastructure is deployed on a set of computing nodes provided by the Supercomputer Center of South Ural State University. The MSVF is being implemented as a separate Java-based microservice with its own web and REST interface, providing integration and execution of validation methods and objectives.
To support the validation process, a set of standard validation methods is being developed and integrated to the MSVF, including component-level, integration-level, load and stability testing methods. 5
In this paper, we provided an overview of existing validation approaches that used to test distributed and cloud systems. On the basis of this overview, we presented a model of validation of microservice systems, supporting such microservice features as open interface, containerization and autonomy. This methodology covers the microservice systems development process, from the creation of a separate microservice to the production environment continuous stability testing. Based on this model, an architecture of microservice systems validation framework was presented.
Currently we implement the microservice systems validation framework as a software solution that would support microservice applications validation in the cloud environment. It would provide its users a catalog of microservice applications testing methods and an API for integration with continuous integration platforms. Currently, the microservice systems validation framework is being developed to support VAMP and Mjolnirr microservice platforms and Jenkins as a continuous integration framework.
The reported study was partially supported by RFBR, research project No. 14-07-00420-a and by Grant of the President of the Russian Federation No. MK7524.2015.9.