@fristi @happymoomoo @freemo Have you considered using databases specifically geared towards in-memory retrieval? Redis is a really good choice. You can always use both, Redis for the fast reads/cache and then have a Postgres layer that keeps what you want to persist to disk.
@fristi
Aside from the obvious risk of loosing your data postgres on ramdisk works fine. But only use it for volatile data
I doubt its a good choice if all you want is to improve performance on a normal persistant db.
@fristi
Sounds like a solution that shoukd exist in plorma ideally but doesnt. As someone mentioned using inmemory redis cache along side a database db would be how pleorma should have designed that.
With that said you can potentially improve things by tweaking the write cache in postgres however if you avoid the db you still risk volatility issues... The less that stays in memory before writing to db the less your reliability. Read cacheing is much safer but doesnt prevent your hard drive from being hit with a write.
@fristi
Good luck, hit me up if i can help
@djsumdog @happymoomoo