Hi Jihoon, I would say that 5 sec could be actually a pretty good result at such. Did you have a chance to check which exact rules contributed to the planning time? You may inject a listener to VolcanoPlanner to check that.
Regards, Vladimir Вт, 9 марта 2021 г. в 05:37, Jihoon Son <[email protected]>: > Hi all, > > I posted the same question on the ASF slack channel, but am posting > here as well to get a quicker response. > > I'm seeing an issue in query planning that it takes a long time (+5 > sec) for a giant union query that has 120 subqueries in it. I captured > a flame graph (attached in this email) to see where the bottleneck is, > and based on the flame graph, I believe the query planner spent most > of time to explore the search space of candidate plans to find the > best plan. This seems because of those many subqueries in the same > union query. Is my understanding correct? If so, for this particular > case, it seems possible to parallelize exploring the search space. Do > you have any plan for parallelizing this part? I'm not sure whether > it's already done though in the master branch. I tried to search for a > jira ticket on https://issues.apache.org/jira/browse/CALCITE, but > couldn't find anything with my search skill. > > Thanks, > Jihoon >
