YSQL: Reduce execution time of OpenTable by table_id #10717
Comments
d-uspenskiy
added
the
kind/enhancement
d-uspenskiy
changed the title
d-uspenskiy
added a commit
that referenced
this issue
Dec 15, 2021
Summary: client's `OpenTable` function calls 2 remote procedures on master: - GetTableSchema - GetTableLocations These functions are called sequentially. It is possible to call them simultaneously in case `table_id` is known. This will help to reduce `OpenTable` execution time from `490-515` ms to `328-335` ms (tested on AWS `c5d.large` RF=3 cluster. Where t-server was running in `US West` and master in `Asia Pacific` regions). Actually execution time can be reduced to approximately `160`ms (RTT on distributed cluster) by disabling TCP slow start alghoritm. Test Plan: Jenkins Reviewers: alex, dsrinivasan, sergei Reviewed By: sergei Subscribers: yql Differential Revision: https://phabricator.dev.yugabyte.com/D14133
d-uspenskiy
added a commit
that referenced
this issue
Dec 16, 2021
Summary: In the previous diff https://phabricator.dev.yugabyte.com/D14133 one of `Format`'s argument has been removed, but format string has not been changed accordingly. Test Plan: Jankins Reviewers: alex Reviewed By: alex Subscribers: yql Differential Revision: https://phabricator.dev.yugabyte.com/D14398
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
client's
OpenTablefunction calls 2 remote procedures on master:These functions are called sequentially. It is possible to call them simultaneously in case
table_idis known.This will help to reduce
OpenTableexecution time from490-515ms to328-335ms (tested onc5d.large RF=3cluster with master in different region).The text was updated successfully, but these errors were encountered: