Is it a bad idea to run a TypeSense instance per index? I'm dealing with large amounts of data and I'm thinking splitting each index on its own instance would help me scale much easier. Are there any downsides to this approach?

If the data is already split across separate collections, then searches and indexing is already well parallelized across collections. The only benefits you'd then get by putting each collection in separate instances are: 1) Typesense process restart times will be faster (since only that single collection has to be reloaded into memory). However, hopefully you're not having to restart the Typesense process all the time - it's meant to be running as a long-running service. 2) If you have a ton of traffic to just one collection, you can scale just that collection's instance up and add more CPU cores, and keep the other instances smaller. The only downside would be the additional infrastructure management overhead. But if you're on Typesense Cloud, we take care of that for you, so this shouldn't be an issue either.

Oh I see, when I said "large amounts of data" I was meaning a lot of data indexed i.e. huge RAM usage, cause one of the indexes will start at requiring about 64 GB of server RAM just on its own However based on what you already written I conclude splitting should not be a problem

Correct