pingcap/tidb

Loose index scan can be used to speed up DISTINCT upon JOIN

Open

#18,693 创建于 2020年7月20日

在 GitHub 查看
 (12 评论) (0 反应) (1 负责人)Go (40,090 star) (6,186 fork)batch import
challenge-programhelp wantedsig/executionsig/planner

描述

Description

In issue https://github.com/pingcap/tidb/issues/14460. It proposed a scan method called loose index scan. This method is to scan the index and skip the same value.

In this issue, we describe a scene that is not mentioned in that issue.

Consider query select distinct t2.a from t1, t2 where t1.a=t2.a.

Since we need unique t2.a. So we also don't need t1.a return the same value twice, which means that we can apply a loose index scan on the t1 side.

Score

6600

Mentor

@winoros (Slack ID: Cui Yiding)

#sig-planner

Recommend Skills

  • Golang and Rust

贡献者指南

技术栈
gorust
领域
database
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
5
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
over 1 week
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
blocked
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
Go programmingDatabase indexingSQL JOINsTiDB internals
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
15
研究方向
Start by examining the referenced issue #14460 which introduced the loose index scan concept. Then study the TiDB source code in the planner and executor packages to understand current DISTINCT and JOIN processing. Review the comments on this issue for any additional guidance from mentor @winoros. The goal is to implement a new scan method that skips duplicate values on the t1 side of the join.