[wmccullough]

I agree completely.

I think that regardless of whether microservices works for anyone or not, they came about to address a real issue that we still have, but that I’m not sure anyone has fully solved.

I think that microservices are an expression of us trying to get to a solution that enables loose coupling, hard isolation of compute based on categorical functions. We wanted a way to keep Bob from the other team from messing with our components.

I think most organizations really need a mixture of monolithic and microservices. If anyone jumps off the cliff with the attitude that one methodology is right or wrong, they deserve the outcome that they get. A lot of the blogs at the time espoused the benefits without bothering to explain that Microservices were perhaps a crescent wrench and really most of the time we needed a pair of pliers.

[tabtab]

The problem is it's not clear when to use what. Some get confused and use it in the wrong place. Here are some questions to ask before you use microservices.

Does the service need an independent and dedicated team to manage its complexities, or is it a "part time" job? Try a Stored Procedure first if its the second.

Is the existing organization structure (command hierarchy) prepared and ready for a dedicated service? (Conway's law) Remember, sharing a service introduces a dependency between all service users. Sharing ain't free.

Do you really have a scalability problem, or have you just not bothered to tune existing processes and queries? Don't scrap a car just because it has a flat tire.