知识卡片
数据布局决定汇总算子
搜索含cat和dog的文档时,若每叶都有完整索引,可把词查询分给不同叶,根取文档集合的交集。若文档分片存储,每叶先对本地数据完成双词查询,根再取各片结果的并集。看似同一搜索需求,数据放置改变了各层工作和正确的汇总规则。
这张卡片看不懂?
这张卡片的反馈已满 10 条,暂时无法继续提交。
已有反馈 · 0 / 10 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- Chapter 8. Scatter/Gather · 查看原书依据
In the sharded case, the root node is responsible for generating the union of all of the documents returned by all of the different shards and returning this complete set of documents back up to the user. In Figure 8-3, the first leaf serves documents 1
参考来源
- 位置:《Designing Distributed Systems, 2nd Edition》第08章《Chapter 8. Scatter/Gather》(来源项目 book-sources;源文件:extracted/Designing Distributed Systems, 2nd Edition/_epub-src/OEBPS/ch08.html)
- 结论依据:原文两种文档搜索例子分别展示交集和并集,说明汇总必须服从分解方式。
- 原始内容:In the sharded case, the root node is responsible for generating the union of all of the documents returned by all of the different shards and returning this complete set of documents back up to the user. In Figure 8-3, the first leaf serves documents 1