Hi, Xia This FLIP looks good to me, +1.
I've two questions: 1. >> Accordingly, in terms of implementation, we will delay the codegen and creation of the join operators until runtime. How are you delaying codegen to runtime, the current runtime is not SQL planner aware. in other words, how do I understand this sentence? 2. FLIP-469 mentions passing StreamGraphOptimizationStrategy to runtime via option `execution.batch.adaptive.stream-graph-optimization.strategies`. In SQL planner if you have multiple different optimization strategies like broadcast join, skew join, etc... When did you configure these optimization strategies uniformly into `execution.batch.adaptive.stream-graph-optimization.strategies`? Zhu Zhu <reed...@gmail.com> 于2024年7月19日周五 17:41写道： > +1 for the FLIP > > It's a good start to adaptively optimize the logical execution plan with > runtime information. > > Thanks, > Zhu > > Xia Sun <xingbe...@gmail.com> 于2024年7月18日周四 18:23写道： > > > Hi devs, > > > > Junrui Lee, Lei Yang, and I would like to initiate a discussion about > > FLIP-470: Support Adaptive Broadcast Join[1]. > > > > In general, Broadcast Hash Join is currently the most efficient join > > strategy available in Flink. However, its prerequisite is that the input > > data on one side must be sufficiently small; otherwise, it may lead to > > memory overuse or other issues. Currently, due to the lack of precise > > statistics, it is difficult to make accurate estimations during the Flink > > SQL Planning phase. For example, when an upstream Filter operator is > > present, it is easy to overestimate the size of the table, whereas with > > an expansion operator, the table size tends to be underestimated. > Moreover, > > once the join operator is determined, it cannot be modified at runtime. > > > > To address this issue, we plan to introduce Adaptive Broadcast Join > > capability based on FLIP-468: Introducing StreamGraph-Based Job > > Submission[2] > > and FLIP-469: Supports Adaptive Optimization of StreamGraph[3]. This will > > allow the join operator to be dynamically optimized to Broadcast Join > based > > on the actual input data volume at runtime and fallback when the > > optimization > > conditions are not met. > > > > For more details, please refer to FLIP-470[1]. We look forward to your > > feedback. > > > > Best, > > Junrui Lee, Lei Yang and Xia Sun > > > > [1] > > > > > https://cwiki.apache.org/confluence/display/FLINK/FLIP-470%3A+Support+Adaptive+Broadcast+Join > > [2] > > > > > https://cwiki.apache.org/confluence/display/FLINK/FLIP-468%3A+Introducing+StreamGraph-Based+Job+Submission > > [3] > > > > > https://cwiki.apache.org/confluence/display/FLINK/FLIP-469%3A+Supports+Adaptive+Optimization+of+StreamGraph > > >
