Towards Automatic Migration to Microservices

Distributed Adaptive Systems (DAS) Unit
A Model-Driven Approach Towards
Automatic Migration to Microservices
Antonio Bucchiarone, Kemal Soysal, Claudio Guidi
bucchiarone@fbk.eu, kemal.Soysal@ls-it-solutions.de, cguidi@italianasoftware.com
LS IT-Solutions GmbH

Microservices 2020 Towards Automatic Migration to Microservices 2
Introduction
 Scalability and Availability of modern software systems.
 Microservices is an architectural style originating from SOA.
 Microservices to realize complex systems.
 A set of small services that operate independently.
 Services that are easy to maintain and evolve.
 Each service represents a single business capability.
 Each service can be developed and managed by a single team.
 Migration from monolithic applications to microservices-based
applications.
 Is a challenging task that very often is done manually and without a
supporting tool.

The microservice paradigm still misses a conceptual
model able to support engineers starting from early
phases of development.
Research Challenge
conceptual model
A Model-Driven approach for
the automatic migration to microservices.

Model-Driven Engineering
 Software Engineering Methodology that proposes to shift the
focus of the development from coding to modelling.
 The goal is to reduce the complexity of software development by
raising the level of abstraction:
 Analyzing application properties earlier.
 Introducing automation in the development process.
 Models are expected to allow domain experts to reason about a
certain solution by means of concerns closer to their area of
expertise than to implementation details.
 Model transformation techniques manipulate models to evaluate
the attributes of the application and generate implementation code.

Domain Specific Languages (DSLs)
 Languages introduced for expressing problems by using terms
closer to a particular domain of application.
 The definition of DSLs can be challenging due:
 The nature of the specific domain.
 How the concepts should be interconnected to ease the modelling
activity.
 What kind of concrete syntaxes the users desire to exploits.
 Etc…
 To support the development of DSLs, it is common practice to use a
language workbench.
 A toolkit supporting the definition of various aspects of the DSL under
development
 Syntax, semantics, validation constraints, generators
 Code competition, syntax coloring, find usages, refactoring, etc..

JetBrains MPS - I
 Text-based meta-programming system that enables language-
oriented programming.
 Open Source and used to implement languages with different
notations.
 Based on MPS BaseLanguage is possible to define new custom
languages through extension and composition of concepts.
 A new language is composed by different aspects making its
specification modular and therefore easy to maintain.
 Structure Definition: Abstract Syntax Tree (AST) of a language as a
collection of concepts.
https://www.jetbrains.com/mps/

JetBrains MPS - II
 Structure Definition aspect: Abstract Syntax Tree (AST) of a language as
a collection of concepts.
 Editor Definition aspect deals with the definition of the concrete
syntax for a DSL:
 It specifies the notation (i.e., tabular, diagram, tree, etc..) and the
interaction behavior of the editor.
 Generators Definition aspect is used to define the denotational
semantics for the language concepts.
 AST to text (model-to-text)
 AST to AST (model-to-model)
 Type System Definition and Constraints Definition aspects.
https://www.jetbrains.com/mps/

Jolie Language for Microservices
 A programming language which offers a native linguistic tool for
defining microservices following a structured service-oriented
paradigm.
 It provides an integrated syntax for defining API interfaces and types;
 It provides specific communication primitives for dealing with
communication (synch, asynch).
 It allows for defining the service behaviour in a workflow manner:
easy definition of orchestrators and coordinators of services.
https://www.jolie-lang.org/

 Each operation comes with a request message and a response
message (synchronous operations).

 Declarative Part where all the
interfaces and ports are
defined.
 Behavioral Part where the
implementation of the
operations is provided.
 The primitive execution defines
the execution modality of the
service (concurrent, sequential,
single).

Deploying a Jolie Service
 The available image jolielang/jolie provides a core layer where both
Java and Jolie are installed.
 Docker file allows for the creation of a docker image of the service
“calculator”.

Our Vision towards an Automatic Migration
 There is an extreme and increasingly emerging need to migrate applications
from monolithic to microservices.
 Rare are the approaches that try to make this process automatic and tool
supported.
 Most of the migration approaches proposed are guided by the developers
experience and are not supported by a specific tool or language.
Towards the realization of a
general framework
for automatic migration to microservices.

Model-Driven Migration Approach
 Two components, Two DSLs, and a set of Generators

The Microservice Language - I

Calculator Microservice Model

The Deployment Language
Concepts of the
Deployment
Language
Dockerfile Editor
and Model

Microservices Miner
 To analyze the monolithic application written in Java and to retrieve the set of
needed microservices.
 To search in the Abstract Syntax Tree (AST) of the imported Java code for patterns of
interest and suggest to the developer the set of microservices for the migration.

Microservices “Searcher”
 Each searcher is realized by implementing the interface or the abstract
implementation and finds nodes in the AST that comply to the semantic
understanding of the searcher.

Microservices Generator
 AST to Text (model-to-text) transformation.
 To run a microservice we provide one file for the service specification (.ol)
and one file for each interface the microservices uses (.iol)

Dockerfile Generator
We generate the Dockerfile used to deploy the microservice in a Docker
container.
 This generator is used to generate the corresponding Dockerfile document
needed to deploy the microservice in a Docker container.

Prototype Implementation
https://github.com/antbucc/Migration

Open Research Challenges
• Microservices Modeling
• Define a more abstract DSL that can be used to model more general
microservices.
• Microservices Mining
• AI-based approach to extract automatically microservices from
monolithic application
• Microservices Composition
• Introduce the way to realize not only microservices but a complete
Microservice-based Application (MSA)
• Abstraction / Generalization
• To target multiple microservices technologies from the same abstraction

Distributed Adaptive Systems (DAS) Unit
A Model-Driven Approach Towards
Automatic Migration to Microservices
Antonio Bucchiarone
bucchiarone@fbk.eu
Thank you!
Any questions?

Towards Automatic Migration to Microservices

More Related Content

Similar to Towards Automatic Migration to Microservices

Recently uploaded

Towards Automatic Migration to Microservices