3 ms·
I’d add that everything is a micro service - usually that is - which simplifies the process of contributing.
by PartiallyTyped 4y ago
I’d add that everything is a micro service - usually that is - which simplifies the process of contributing.
- papito 4y agoQuite the opposite. I've fixed bugs on my second day on a job when it was a simple, testable monolith - the way it SHOULD be for most companies still. In a distributed system, a small change can have a giant blast radius. You don't know if the service you are touching was part of resilience testing. There are hundreds, it's easy to miss. What if you make this microservice throw all the time? What happens? Does it bring down anything else? I don't understand why people still think that microservices make things easier. Distributed systems have always been insanely complex to reason about, and that never changed because Node or Docker, or something. He even complained that it was impossible to get anything working locally because everything was so interconnected and complicated, so there is your answer.
- PartiallyTyped 4y agoThose are all good points that I hadn't considered. The idea behind microservices is separation of concerns with horizontal scalability. The microservices should be independent enough that it's possible to test the integration in stages without bringing the whole system down. My team relies on services offered by cloud providers, we are just composing them together, which means that we don't have to create userpools for example.