Hi there, I have a few questions: For context, we ...
# ask-questions
a
Hi there, I have a few questions: For context, we have hosted GrowthBook on our own EC2 machine. • Since all requests are handled by a single instance, this is a blocker to scaling. How can we separate the frontend and backend so that I can set up an auto-scaling group for the backend? • Is there a way to rotate Docker logs? For example:
/var/lib/docker/containers/015ae01b454a1f06e7a6f6ed77b4948a68523579916da749a5a9c9144a0d009b/015ae01b454a1f06e7a6f6ed77b4948a68523579916da749a5a9c9144a0d009b-json.log
This log file was close to 6 GB. ^ • I’ve noticed that GrowthBook does not maintain event data, which is why
on_experiment_viewed
is called multiple times for different threads. This is an open question: how has everyone handled maintaining deduplication?
f
For scaling, GrowthBook's backend (API) and frontend are stateless. you can add as many as you want- the data store is mongo/documentdb.
You can probably scale to a huge volume by just putting a CDN infront of the SDK endpoint request
we de-dup the events on the SQL side, so its not a an issue to fire it multiple times (in terms of results). If there are efficiency reasons for blocking additional calls, you're welcome to add some state to see if events have been fired