Skip to content

Latest commit

 

History

History
25 lines (16 loc) · 2.57 KB

MicroservicesLinks.md

File metadata and controls

25 lines (16 loc) · 2.57 KB

Links on Microservices

Microservices have a dual nature: On the one hand, they give independence of code, testing, deploy times, versions, tools, techniques, etc. The ability for a team to work on and deploy an isolated unit of functionality when ready is great. It really is an architectural style for the agile and Continuous Deployment era.

On the other hand, Microservices imposes a strong contract with how a service behaves. That service will have to be a "good citizen" of the ecosystem, even while it has freedom of how it implements these requirements. For instance, it will need to participate in centralised logging and monitoring systems. It will need health checks for load balancers. Those load balancers can create and terminate instances at will, so work-sharing is a concern, as is graceful shutdown. Coordinating all this across many teams' code is different to doing it to one monlithic app.

Because of the overhead, without scale you don't get much benefit from Microservices.

Links on shared code in microservices

There's a point repeated here, that Dont Repeat Yourself is an excellent rule, but it is not the only rule and it is contextual. Beware of taking DRY too far; especially at larger scales where it breaks down, e.g. across the boundaries of one executable. You remove duplication but you create coupling and the costs can outweigh the benefits. Duplication is better than the wrong abstraction.

Microservices, of course, creates lots of these boundaries. Be wary of code that is required to make these services interoperate and is not common enough to form part of the language framework or an open-source public package. Microservices should be mostly independent.