Large sparse table cause apply snapshot slow
Nobody has claimed this yet.
- Dominant language
- C++
- Stars
- 1k
- Forks
- 423
- Avg merge
- 1d 15h
- Merged PRs (30d)
- 24
Description
Enhancement
A sparse table with average row size~15.17 bytes.
Because we set region-split-keys = 50000000 and region-split-size = "500MiB" on serverless tier, its regions snapshot contains more than 54,000,000 rows.
Decoding the key-values of the snapshot takes 220 seconds and ingest it into Delta-Tree cost 17 seconds.
The ingest phase is executed by only 1 thread, these slow ingest become the bottleneck when we adding TiFlash replica for a large amount of Regions cause regions apply snapshot timeout.
After changing following configs, the ingest cost down to less than 4 seconds.
[profiles]
[profiles.default]
dt_segment_limit_rows = 5000000
dt_segment_force_merge_delta_rows = 671088640
dt_segment_stop_write_delta_rows = 1342177280
We need more tests to ensure whether it can become the default config values in the master branch.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by locating the profile settings for dt_segment_limit_rows, dt_segment_force_merge_delta_rows, and dt_segment_stop_write_delta_rows, then find tests covering sparse-table snapshot ingest. Compare the current defaults with the reported values and measure ingest time for large snapshots. Done means tests justify whether these values can become the master-branch defaults.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- databases, performance, testing
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100