Re: Updates on Calcite benchmark development
I think I shared Google Doc link on JIRA. If not, I will share a link.
One question we have is what other Calcite supported data sources, in
addition to JDBC-based ones, are worth benchmarking.
Worth here would be determined by a) utility/demand for the data source,
and b) maturity of the Calcite adapter.
Please share your thoughts.
On Friday, April 27, 2018, Michael Mior <mmior@xxxxxxxxxxxx> wrote:
> Great to hear! Thanks for the update and keep us posted if there's any way
> the rest of us can help.
> Michael Mior
> Le jeu. 26 avr. 2018 à 22:51, Edmon Begoli <ebegoli@xxxxxxxxx> a écrit :
> > Dear all,
> > My colleagues and I are completing the implementation of the first
> > benchmark suite for Calcite, with the expectation to have something in
> > place in the next 2-4 weeks, and available for testing.
> > The plan is to make it available for your review and testing, and then
> > release it
> > Current status is:
> > We have 80+ out of 99 TPC-DS queries made compatible to Calcite for
> > Postgres backend. (Calcite-178)
> > It will be compatible with most of the backends, but the queries will
> > testing and validation.
> > We need to complete following two tasks to complete the initial benchmark
> > work:
> > 1) an accurate method to measure and instrument the Calcite performance
> > since the time measurement for Calcite may involve overhead associated
> > Avatica/JDBC.
> > 2) fix queries that require syntax modification for Calcite syntax or to
> > account for some lack of support
> > We are not sure yet how bad the 2 part might be. It might involve some
> > patches or workarounds, which also might vary from backend to backend.
> > Big thanks to my R&D team members, Dr. Seung-Hwan Lim and Ashwin Vajantri
> > for working on this, and specially on tedious TPC-DS query syntax fixing
> > make it work with Calcite.