search

t3-innovation-network / OCF-Collab

0 stars 0 forks source link

Problem of flooding the RB/PNA with requests #17

Open stuartasutton opened 3 years ago

stuartasutton commented 3 years ago

@neaf & @jbaird123 & @jeannekitchens, in closed issue #14, you stated, Tomasz, that there was likely an elasticsearch issue caused by "flooding the RB/PNA with requests". Given how our current environment works--i.e., one off requests for a framework--I don't see how this is possible. Would this be a problem when going to scale where there may in fact be lots of requests through the RB (less likely the PNA)?

neaf commented 3 years ago

It isn't related to ElasticSearch in any way. Just a concurrency limit on the services.

Right now it's deployed using Cloud66 service which automates the infrastructure setup. I don't think the server they use actually supports concurrency without Enterprise version.

In the end I don't believe it's a proper solution for production-grade deployment. The infrastructure should be reproducible (using infra-as-a-config like Terraform), with proper redundancy, resource management, load balancers, back-ups and probably dozens of other things I don't know about as dev-ops is not my area of expertise.

I've mentioned this already to @jbaird123. We should be fine for the proof-of-concept phase. The production environment has a little bigger servers and I didn't encounter the issue (I'll stress test it though to fully understand).

However, at some point it should be handled by someone more experienced with systems deployment. I could figure it out but that would mean spending undetermined time.

stuartasutton commented 3 years ago

@neaf, @jbaird123, @jeannekitchens, Tomasz message above needs to be condensed into a sentence or two and listed as a Phase 3 "ask" in the OCFC budget table at row 12.