Riza Suminto has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/20612 )

Change subject: IMPALA-3825: Delegate runtime filter aggregation to some 
executors
......................................................................


Patch Set 8:

(3 comments)

http://gerrit.cloudera.org:8080/#/c/20612/7//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/20612/7//COMMIT_MSG@20
PS7, Line 20: Query option MAX_NUM_FILTERS_AGGREGATED_PER_HOST is added
            : to control this feature. Given N as the number of backend 
executors
            : excluding the coordinator, the selected number of intermediate
            : aggregators M = ceil(N / MAX_NUM_FILTERS_AGGREGATED_PER_HOST). 
Setting
            : MAX_NUM_FILTERS_AGGREGATED_PER_HOST <= 1 will disable
> MAX_NUM_FILTER_AGGREGATOR seems ok for me in the short term to test this fe
Done. Changed MAX_NUM_FILTER_AGGREGATOR to MAX_NUM_FILTERS_AGGREGATED_PER_HOST.


http://gerrit.cloudera.org:8080/#/c/20612/7/be/src/scheduling/scheduler.cc
File be/src/scheduling/scheduler.cc:

http://gerrit.cloudera.org:8080/#/c/20612/7/be/src/scheduling/scheduler.cc@282
PS7, Line 282: typedef vector<pair<int, int>> InstanceToAggPairs;
             :
             : v
> Can you move this to network-util.h?
Done


http://gerrit.cloudera.org:8080/#/c/20612/7/be/src/scheduling/scheduler.cc@317
PS7, Line 317:     // Put coordinator grou
> Does num_agg = 1 make sense?
I think it still make sense.

Say a query has 50 bloom runtime filters and running in 100 executors cluster.
num_agg=1 means coordinator handle only 50 finalized filter instead of 5000 
filter updates.
In this case, coordinator will only responsible to publish the finalized filter 
to all executors, while the aggregation happen in other nodes. Note that only 
coordinator has full KRPC addresses of cluster members via 
Coordinator::BackendState.



--
To view, visit http://gerrit.cloudera.org:8080/20612
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I11d38ed0f223d6e5b32a19ebe725af7738ee4ab0
Gerrit-Change-Number: 20612
Gerrit-PatchSet: 8
Gerrit-Owner: Riza Suminto <riza.sumi...@cloudera.com>
Gerrit-Reviewer: Abhishek Rawat <ara...@cloudera.com>
Gerrit-Reviewer: Csaba Ringhofer <csringho...@cloudera.com>
Gerrit-Reviewer: Impala Public Jenkins <impala-public-jenk...@cloudera.com>
Gerrit-Reviewer: Kurt Deschler <kdesc...@cloudera.com>
Gerrit-Reviewer: Michael Smith <michael.sm...@cloudera.com>
Gerrit-Reviewer: Riza Suminto <riza.sumi...@cloudera.com>
Gerrit-Comment-Date: Wed, 01 Nov 2023 20:19:56 +0000
Gerrit-HasComments: Yes

Reply via email to