Content
@
https://warpcast.com/~/channel/uno
0 reply
0 recast
0 reaction
christopher
@christopher
We are rebuilding some feed caches and indices. Uno will be slow for a bit while this processes across replicas. Root cause is that using the JOIN/IN keyword for queries is much slower than EXISTS now that we are indexing millions and millions of rows of network data.
3 replies
1 recast
7 reactions
Mahatma Gandhi
@whiteshadow.eth
ok
0 reply
0 recast
0 reaction
higgins.base.eth
@higgins
Under big data scenarios, avoid using `JOIN`/`IN`; use `EXISTS` appropriately, and try to split long arrays into smaller parts to perform concurrent queries whenever possible.
0 reply
0 recast
0 reaction
Luciano
@luciano
🙌
0 reply
0 recast
0 reaction
