←back to thread

MVCC – the part of PostgreSQL we hate the most (2023)

(www.cs.cmu.edu)
366 points virtualwhys | 2 comments | 20 Oct 24 15:30 UTC | HN request time: 0s | source
Show context
WuxiFingerHold ◴[21 Oct 24 04:11 UTC] No.41900626[source]▶
Putting aside the license / closed source issues with CockroachDB (CRDB) and just focus at it technically: CRDB uses MVVM too, but its storage is a key-value store. I know it uses some kind of garbage collection to remove the old versions.

I wonder if CRDB (or other newer designed DBs) has circumvented those issues? Or don't we just hear from those issues as CRDB and the other newer DBs are not that widely used and mainly in the commercial space?

replies(2): >>41901017 #>>41901777 #
ddorian43 ◴[21 Oct 24 08:05 UTC] No.41901777[source]▶
This is fixed in YugabyteDB that reuses the PostgreSQL query layer source code but uses it's own storage: https://www.yugabyte.com/blog/improve-postgresql/ (other issues too like XID wraparound etc).

It's also apache2 license.

replies(1): >>41907805 #
1. knowitnone ◴[21 Oct 24 19:48 UTC] No.41907805[source]▶
Thanks for pointing out Yugabyte. I did a search for performance comparisons and there seems (seemed?) to be some performance issues at least back in 2023 https://github.com/yugabyte/yugabyte-db/issues/10108

hopeful performance has improved since

replies(1): >>41911154 #
2. ddorian43 ◴[22 Oct 24 04:19 UTC] No.41911154[source]▶
Yes, performance is much better. But it's a bit apples to oranges comparisons because of distributed/sharding nature (example with bitmap scans https://www.yugabyte.com/blog/bitmap-scans-on-distributed-po...), and always synchronous replication (https://www.yugabyte.com/blog/yugabytedb-resiliency-vs-postg...).