TimescaleDB does not use LSM + SST, it uses B-trees. The writes/reads are still fast as the part that we are usually interested in is the latest chunk, for which the index tree can be loaded into memory.
So is each chunk table like its own file on the disk, how is that part stored? Like given a time query and a sensor value, how do we locate the appropriate chunk table"?
One of the advantages that was in the old videos on design systems is that in some videos there are links to recommended resources for studying.
Fair - when I have time I will eventually need to paste all the resources in 2.0
The audio is very very low.
Ah shit sorry about that I'll have it fixed for next time
Waiting for a collab my man. 🙌
would this be useful for storing historical data for stock prices where we need to aggregate for ex 1m,5m,1hr etc
For sure
Do you have a discord / telegram / linkedin id?
www.linkedin.com/in/jordan-epstein-69b017177
TimescaleDB does not use LSM + SST, it uses B-trees. The writes/reads are still fast as the part that we are usually interested in is the latest chunk, for which the index tree can be loaded into memory.
Oops if I said that, agreed since it forks postgres iirc
Nice content, Jordan.
Keep it coming.
So is each chunk table like its own file on the disk, how is that part stored? Like given a time query and a sensor value, how do we locate the appropriate chunk table"?
I imagine a separate file path yeah.
Thanks
need to wear a mask like MF doom
Yeah it's a shame I'm in too deep now