[rjkennedy98]

It is actually sold as a way to fix bad engineering? I've literally never heard that before. I think almost everyone knows that microservices are hard.

On the other hand they are often sold as a way to increase developer velocity. And I do sometimes wonder if that is the case (based on personal experience).

[mrits]

We did a massive rewrite of a 7 year old giant code base to microservices gradually over a couple years. One of the only few positives that we gained from it was that when we crashed we didn't bring down everything at once. The biggest negative was that we ended up with an extremely noisy event bus that made backups and diagnosing out of sync issues very problematic.

[cc81]

Yes, it is a pretty common saying that it forces separation of concerns and avoids the spaghetti that monoliths end up like.

People are selling it as easy. You can deploy them independently, each service is small with low complexity.

That it is not that easy in reality and that the total complexity increase is so big is not told as often. There is a team at my work with 6 developers developing something that processes a couple of gigs of data each day and has a front end serving maybe a hundred users total, I would guess 20 would be logged in at the same time.

Kubernetes, 10 microservices, 2 different types of databases.

[bcrosby95]

Yes. I see people push microservices all the time because their monolith is poorly designed. If your monolith is messy, it's on you, not monoliths.