Beyond hosted DBs - infrastructure provisioning/integration pain-points

[tessier0ashpool]

based on the discussion in the Discord channel and forums, setting up a hosted database is a pain-point for a bunch of people

I’ll start working on some quick tutorials for setting up RDS and S3 for use with RedwoodJS (thanks for the idea, @dom!), kind of like the really nice Netlify one here in the How-To section:

Connecting Netlify CMS with RedwoodJS

@rodocite Is this a hint on what’s to come? ^^ not that I am aware of. But I’m wishing + planting seeds The current inflexibility of RedwoodJS folder structure (I wanted to make a monorepo with all the RedwoodJS /web apps in it sharing the same /api). ^^ Hmm, did you know that a given Redwood Project (i.e. App) can be either a Web side or API side, and doesn’t even require a DB? So one thing you could do is create a project that’s only the API. Then spin it up with yarn rw dev…

beyond this, i did want to ask about what other Infrastructure provisioning and integration pain-points exist for those of you that do not spend your lives in the AWS console like an ogre under a bridge
based on my more general and very scientific survey of front-end / fullstack devs:

• job runners / ETL - there are so many to choose from here, but I can’t find a clear winner or an “easy button” hosted version. i would LOVE to be wrong. please prove me wrong.

• “microservices” - probably flask backed with apis/datastores and/or AWS lambdas
  – really, this is for anything that’s too big, complex, runs on a schedule, or needs more logging/monitoring than a netlify function.
  – maybe we’d want to also have a generator for the service / stubs?

also: I’ve not heard much about S3, but since it’s commonly used & easy-but-repetitive, it seems like it should be addressed.

finally - the idea here would be to generate the basic infrastructure that "just works"™ with sane defaults. it wouldn’t be to make something super configurable or customizable.
I do need to think through the design quite a bite, but for right now, this information would be super useful by itself. Thanks!

[hd10]

Thanks for starting this discussion.

@tessier0ashpool

“microservices” - probably flask backed with apis/datastores and/or AWS lambdas
– really, this is for anything that’s too big, complex, runs on a schedule, or needs more logging/monitoring than a netlify function.

For microservices or for the API in general, using managed container solutions would be pretty great. Something like Cloud Run or AWS Fargate. That would also handle the connection issues for DBs. I think you could just create an Apollo Express Server instead of the createGraphQLHandler for the API side and turn the api folder into a docker container.