pingcap/tidb

Accelerate FindColumn when table has huge number of columns

Open

#11,151 opened on 2019年7月9日

GitHub で見る
 (3 comments) (0 reactions) (0 assignees)Go (40,090 stars) (6,186 forks)batch import
help wantedtype/enhancementtype/performance

説明

Feature Request

Is your feature request related to a problem? Please describe:

will be frequently called when build query plan or build insert value list.

now it uses linear search to find column info match target name....

when the table has limitation columns it works well...

but when table got many columns (e.g. 200+ columns), upper logic will find 200 columns to do linear search in 200 items.

image

Describe the feature you'd like:

maybe we should adaptively replace a linear search with another method when table has many columns

コントリビューターガイド

技術スタック
go
領域
backenddatabaseperformance
Issue 種別
feature
難度新しい貢献者にとっての実装難度です。1 は非常に小さな変更、5 は専門的な作業です。
3
推定時間経験ある貢献者が調査、実装、テスト、pull request 準備にかけるおおよその時間範囲です。
half day
活動状況Issue が今どれだけ取り組みやすいかを示します。新しい、活発、古い、ブロック中、またはメンテナー入力待ちなどです。
stale
明確さ期待される変更、受け入れ条件、次の手順がどれだけ明確かを示します。
clear
前提条件
Go programmingTiDB internalsalgorithm optimizationperformance profiling
初心者向け度初回貢献者にどれだけ取り組みやすいかを 1-100 で推定したスコアです。
45
調査方針
The issue points to two functions: FindCols in table/column.go and FindColumn in expression/schema.go. Both currently use linear search to find column info by name. The goal is to replace linear search with a more efficient method (e.g., hash map) when the number of columns is large. Research should start by understanding the current implementation and the data structures used. Then, consider implementing an adaptive approach: keep linear search for small column counts and switch to a map or sorted slice with binary search for larger counts. Profile the performance impact on query planning and insert value building. Check the three comments for any additional insights or suggestions.