Skip to main content

Microservices


On today’s blog entry we are going to talk about the Microservices architecture. This architecture has an approach where when we are developing a single app, we develop it as lots of small services, each of them running on its own process and communicating with each other with something that can bind them together. Now I have already used microservices without knowing I was actually using them, perhaps this architecture is quite intuitive for best practices or something like that, but anyway the author of the article gives some characteristics about this which are the following:

·         Componentization via Services: The components of the microservice need to be gather into libraries that will be used later on by the microservice to communicate with remote parts of the whole program.
·         Organized around Business Capabilities: This characteristic refers that we need to create divisions, each of the divisions need to be focused on the business capability which include various things like storage, interface, and many other stuff, now for this part the team developing this part needs to have different knowledge not only about programming, but also knowledge about project management, databases and obviously user interaction or experience.
·         Products not Projects The resultant software needs to be acknowledged not as another project, but as a potential product which characteristic can be used to improve the whole capability of the business than just a small part of it
·         Smart endpoints and dumb pipes: The endpoints should be for all the logic of receiving requests and applying it accordingly, and the pipes should be like small messages buses that behaves as a message router.
·         Decentralized Governance: This refers that the developers should use any tool at their disposal so the work can be done more faster and in form, but also they should probably code in any language so they can use those tools at their fullest so the system is not language based, but technology based.
·         Decentralized Data Management: the conceptual model of the world will differ between systems, so you should prefer letting each service manage its own database, either different instances of the same database technology (Polyglot Persistence).
·         Infrastructure Automation: automated test ad automate deployment.
·         Design for failure: Our software must tolerate the failure of the external services.
·         Evolutionary Design: Design software that is capable to evolve in the future.

Comments

Post a Comment

Popular posts from this blog

Is Design Dead?

On today’s entry we are going to talk about the article “Is design dead?”, which talks about how the design has been changing through the years and how the good practices were developed in order to create a good design. Is no lie that for all the people that develops software, is kind of difficult to come up with a good design because of the lack of and standards to create it, other areas may have standards or even a methodology, but not us. Because of these needs, developers have created a bunch of good practices, not to create, but to reach a good design probably in a 85% of success. For example, a good practice is a that the code we create has to be modular in order to make corrections or complete alterations in a easier, this changes may occur for many reasons, manteinance, update the information or just to simply chage the entire code to something more efficient, the same goes to the design, but the problem is that a design is more difficult to modify when is already d...
Post a Comment
Read more

Understanding the SOLID principles

On today’s blog we are going to talk about the SOLID principles, which were discovered by Robert Martin, the famous Uncle Bob from the podcast of the previous entry. This principles are recommended to be used whenever someone wants to create an object oriented program. There are 5 principles where the acronym SOLID comes from which are the following S ingle Responsibility Principle (SRP) -> it says that any class should only have one responsibility. Yet, the author ends up saying that there might be some classes that have two responsibilities (the invoicing class that has some logging routines). In the end, I didn’t understand this principle well because I found the explanation confusing (as the author said, practice is much more difficult than theory).  O pen/Closed Principle (OPC) -> open extension but closed modification Although the author says that if you change a parent class method it might kill another part of the ...
Post a Comment
Read more