perf(mito): scan SSTs and memtables in parallel (#2852)

* feat: seq scan support parallelism

* feat: scan region by parallelism in config

* feat: enlarge channel size

* chore: parallel builder logs

* feat: use parallel reader accroding to source num

* chore: 128 channel size

* feat: add fetch cost metrics

* feat: add channel size to config

* feat: builder cost

* feat: logs

* feat: compiler error

* feat: fetch cost

* feat: convert cost

* chore: Revert "feat: logs"

This reverts commit 01e0df2c3a.

* chore: fix compiler errors

* feat: reduce channel size to 32

* chore: use workspace tokio-stream

* test: test scan in parallel

* chore: comment typos

* refactor: build all sources first

* test: test 0 parallelism

* feat: use parallel scan by default

* docs: update config example

* feat: log parallelism

* refactor: keep config in engine inner

* refactor: rename parallelism method

* docs: update docs

* test: fix config api test

* docs: update parallel scan comment

* feat: 0 for default parallelism
This commit is contained in:
Yingwen
2023-12-11 14:43:17 +08:00
committed by GitHub
parent 178018143d
commit 6a57f4975e
21 changed files with 398 additions and 51 deletions

View File

@@ -157,6 +157,13 @@ vector_cache_size = "512MB"
page_cache_size = "512MB"
# Buffer size for SST writing.
sst_write_buffer_size = "8MB"
# Parallelism to scan a region (default: 1/4 of cpu cores).
# - 0: using the default value (1/4 of cpu cores).
# - 1: scan in current thread.
# - n: scan in parallelism n.
scan_parallelism = 0
# Capacity of the channel to send data from parallel scan tasks to the main task (default 32).
parallel_scan_channel_size = 32
# Log options
# [logging]