将尽可能多的判断更贴近数据源,以使查询时能跳过无关的数据,显著地减少数据传输或计算的开销。
同样的理念发生在软件架构设计上,例如将数据计算和准备工作尽可能交予数据库(如UDF)以及后端完成,进而减少前端或中间件层面定制开发的数据计算工作
If a predicate cannot be pushed down, the query executed on the remote database engine excludes that predicate, so filtering has to be done using Databricks Runtime.
https://docs.pingcap.com/zh/tidb/dev/predicate-push-down
There was an error while loading. Please reload this page.