Comet Native Executors: sample a range exchange without re-running the stages below it
维护者通常 2 天内回复
还没有人认领这个 Issue。
评估
- 难度
- 5/5
- 预计耗时
- 一周以上
- 新手友好度
- 35/100
- Issue 类型
- 功能
- 描述清晰度
- 基本清楚
- 活跃度
- 活跃
调研方向
Start with range_exchange in native/native-executors/src/assembly.rs and collect in src/frontend.rs to trace how sampling runs the stages below an exchange. Compare the proposed approaches, including materialized shuffle reuse and runtime sampling; done means range exchanges can be sampled without repeatedly rerunning lower stages. The issue does not select an approach, so clarify the design before implementation.
由索引模型根据 Issue 内容生成。
描述
What is the problem the feature request solves?
To range-partition an exchange, the Comet scheduler (#7289) runs a sampling job of its own: a SampleExec over the stage that feeds the exchange. That is a separate Ballista job, so it runs that stage and every stage below it, then the query runs them all again. Every further range exchange above them adds another run. Spark's RangePartitioner also re-runs the child, but it reuses the shuffle outputs below it.
The sampling jobs also run one after another while the query is being planned, even for range exchanges whose inputs don't depend on each other.
Describe the potential solution
Options, roughly from least to most work:
- Sample range exchanges whose inputs are independent concurrently.
- Sample from the materialized shuffle output below the exchange, so only the producing stage runs twice.
- Find the split points at run time, as Ballista's adaptive planner does for window functions:
RuntimeStatsExeckeeps a quantile sketch andUnorderedRangeRepartitionExeccuts by it. That covers only the first sort key today, so it would need compound keys. The driver-side sampler,SampleExecand its codec would then go away.
Additional context
The sampling is range_exchange in native/native-executors/src/assembly.rs and collect in src/frontend.rs.
Part of #7288.
Willingness to contribute
Yes. I can contribute this feature independently.
- 主要语言
- Scala
- 星标
- 1.3k
- 派生
- 405
- 平均合并
- 3 天 1 小时
- 30 天内合并 PR
- 277
环境准备
- 没有 Dockerfile 或 Docker Compose 文件
- 有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
apache/datafusion-comet 的其他 Issue
-
area:ci enhancement priority:low requires-triage
难度 2/5 1-3 小时 新手友好度 74/100
apache/datafusion-comet#7319 · 1 条评论 ·
维护者通常 2 天内回复
-
area:ci bug priority:low requires-triage
难度 2/5 1-3 小时 新手友好度 78/100
apache/datafusion-comet#7315 ·
维护者通常 2 天内回复
-
requires-triage
难度 2/5 1-3 小时 新手友好度 72/100
apache/datafusion-comet#7240 ·
维护者通常 2 天内回复
-
bug priority:low
难度 2/5 1-3 小时 新手友好度 82/100
apache/datafusion-comet#7176 ·
维护者通常 2 天内回复
-
docs: update Scala comments that reference the removed iceberg-writes.md page可能已有人在做 @Abhishek-Mohan 于 4 天前认领。 未关闭enhancement
难度 1/5 1-3 小时 新手友好度 83/100
apache/datafusion-comet#7101 ·
维护者通常 2 天内回复
查看 apache/datafusion-comet 的全部 Issue
相似的 Issue
-
难度 2/5 1-3 小时 新手友好度 75/100
riscv-boom/riscv-boom#835 ·
-
Add a base64url Stringable helper可能已有人在做 @nafg 今天认领。 未关闭
难度 2/5 1-3 小时 新手友好度 68/100
nafg/simple-router#362 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 62/100
propensive/soundness#2351 ·
维护者通常 1 天内回复
-
Traffic event delivered/rejected counters appear swapped in TrafficStateController.updateWithReceipt未关闭
难度 2/5 1-3 小时 新手友好度 65/100
digital-asset/canton#705 ·
维护者通常 2 天内回复
-
easy first-contributor first-contributors good first issue help wanted scala up-for-grabs
难度 2/5 1-3 小时 新手友好度 88/100
fork-commit-merge/fork-commit-merge#8457 ·
维护者通常 1 天内回复