Some Questions About Partitioned Raft KV

Translated
1

Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: 关于 partitioned raft kv 的几个小问题

| username: Billmay表妹
  1. Why is reading from partitionkv slow? Also, in what scenarios would mixed transactions be slow?
  2. If partitioned raft kv doesn’t write WAL, how is data safety ensured during a crash?
  3. After introducing partitioned raft kv, how is the block cache of TiKV instance nodes managed?
2
| username: Billmay表妹 | Original post link

Question 1:
image

Additionally, the slowdown is not absolute but depends on the data volume and workload type.

3
| username: tonyxuqqi | Original post link
  1. Partitioned Raft KV generates a lot of MVCC versions that cannot be cleaned up in a timely manner in scenarios with a large number of updates due to less frequent compaction (cleanup can only occur during compaction). Therefore, scanning these MVCC versions during reads can be quite costly. This issue can be greatly mitigated by adjusting the memtable size parameter, but these adjustments can have some negative impact on write performance, so the default parameters have not been changed.
  2. Ensured by Raft log replay.
  3. Unified management of block cache, shared among all RocksDB instances.
4
| username: Billmay表妹 | Original post link

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.