On: 100k TPS over a billion rows: the unreasonable effectiveness of SQLite

[slashdave]

Sure. Now keep everything in memory and use redis or memcache. Easy to get performance if you change the rules.

[koakuma-chan → slashdave]

You can use SQLite for persistence and a hash map as cache. Or just go for Mongo since it's web scale.

[Yodan2025 → slashdave]

yep, then add an AWS worker in-between

[SJC_Hacker → slashdave]

SQLite can also do in memory

[slashdave → SJC_Hacker]

Yeah, very good point. It all comes down to requirements. If you require persistence, then we can start talking about redundancy and backup, and then suddenly this performance metric becomes far less relevant.

[andersmurphy → slashdave]

Backups are to the second with litestream.