iChauster commented on PR #13426: URL: https://github.com/apache/arrow/pull/13426#issuecomment-1171880413
> I think I was mostly curious about differences in density between the left table and the right table(s). For example, a dense left table and a sparse right table or a sparse left table and a dense right table. The left table roughly defines the keyframes so I would expect the density of the left table to be more significant to performance than the density of the right table. Ah yes, we did try this as part of our asymmetric case. The left table definitely impacts performance the most, with clear separations in time. We also observed another interesting property, which suggests that the time frequency of the right tables does not matter as long as they are >= the left hand table time frequency. That is, if the left hand table has a time frequency of 10 minutes, we see an increase in real_time when joining tables with 1d, 1h, 30m, etc, but the time it takes to join with right hand tables with 10m, 5m, 1m, is the same. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
