If you are happy with Heroku and it's cheap enough then stay there. Any specific requirement that's hard to satisfy on Heroku?

I have my own implementation of a persistent event queue and I was wondering if I'd be better off to use Kafka instead. And Kafka provisioning in Heroku costs quite a bit (€1500/month per standard instance) (not sure how it works yet). Also I wanted to run two instances of my own app that would calculate the same things so that once one goes down there is its backup online. And those would cost €25 per instance. And then if I'd want to share some load and partition my instances (these doubled CPU instances + web) then it would multiply even more. So quite quickly it can become something unwieldy. I'm trying to figure out how my single-thread model scales.

Sounds like you're worrying about scale way before you actually have any problems with scale. IMO that makes sense only if you see that definite growth is right around the corner and there will be scaling needs. But even if that's the case, with just 2-4 instances your load probably won't be too high for "dumb" solutions like putting all the events into a PostgreSQL database. Which makes even more sense given that you want to have a simple experience.

I actually do have PostgreSQL and use it to put all events into 😄 . But I wanted to make my instance reboot-safe. So I'll probably simply host two of them on one DB (but then also web uses that DB but they use different tables). And then potentially I'd want to split it later. But yes, I worry too much and I don't know if I'll need it 🙂

That's it! Thank you. For some reason I couldn't even find it in my browser history!

Haha np! A while back I enjoyed solving a few in clojure. A lot of the collections functions work pretty well for processing dna sequences

I learned my first clojure chops solving project Rosalind problems. https://github.com/tmoerman/rosalind disclaimer: n00b code abound in that repo 😉

I'd love to grow genomics Clojure libraries based on the Clojure data science ones