Ako Suminoe :njp: on Nostr: >It outperformed Postgres querying large time-based datasets. Eg "trending hashtags ...
>It outperformed Postgres querying large time-based datasets. Eg "trending hashtags in the past year".
If you read the bottom of the article, the problem was because Postgres' default config prevents queries from using more than 4MB of memory before it starts using the disk instead, and just bumping that to 16MB was enough to make Postgres outperform Sqlite again.
This kind of feature is important if you are trying to serve hundreds/thousands of concurrent queries while maintaining quality of service.
If you read the bottom of the article, the problem was because Postgres' default config prevents queries from using more than 4MB of memory before it starts using the disk instead, and just bumping that to 16MB was enough to make Postgres outperform Sqlite again.
This kind of feature is important if you are trying to serve hundreds/thousands of concurrent queries while maintaining quality of service.