pingcap/tidb

plan builder is too costly

Open

#27,633 opened on Aug 27, 2021

View on GitHub
 (3 comments) (5 reactions) (2 assignees)Go (40,090 stars) (6,186 forks)batch import
help wantedtype/enhancement

Description

Enhancement

image

There are several apparent part in this picture:

  • generate pseudo statistics cause allocation every time
  • expression rewrite, from ast column name to expression column, the function takes a long time (maybe many columns?)
  • the shadow copy of the column name in expression rewriter

The whole flame graph is, too much allocation caused a lot of GC consumption in the runtime.

image

Maybe the same problem with this issue https://github.com/pingcap/tidb/issues/26868

Contributor guide

Tech stack
gosql
Domain
databaseperformance
Issue type
performance
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
blocked
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Go programmingQuery optimizationProfiling toolsTiDB internals
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
15
Research direction
Examine the flame graphs in the issue to identify hotspots. Look at the plan builder code in TiDB, particularly the pseudo statistics generation, expression rewriting logic, and column name handling. Refer to linked issue #26868 for similar problems. Use Go profiling tools to measure allocations and optimize to reduce GC overhead.