tag:blogger.com,1999:blog-20038672.post5354240951847449738..comments2024-03-19T04:09:53.166-04:00Comments on Robert Haas: Parallel Query v2Robert Haashttp://www.blogger.com/profile/08393677427643988650noreply@blogger.comBlogger10125tag:blogger.com,1999:blog-20038672.post-91458196752077058392017-06-12T05:33:27.100-04:002017-06-12T05:33:27.100-04:00I see that the plan is for Parallel Query with upd...I see that the plan is for Parallel Query with updates planned for Postgres 11. <br /><br />https://wiki.postgresql.org/wiki/Fujitsu_roadmap<br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-20038672.post-7532086091666505702017-05-25T03:43:12.700-04:002017-05-25T03:43:12.700-04:00Great post Robert, keep up the great workGreat post Robert, keep up the great worksugoi desu nehttps://www.blogger.com/profile/11995012278568515245noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-71153573087758464492017-03-16T17:57:54.789-04:002017-03-16T17:57:54.789-04:00Thanks for the response. For ELT workloads a seria...Thanks for the response. For ELT workloads a serial insert after a parallel select would be very handy. In an explain plan most of the time is spent in the query part. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-20038672.post-21007741854866612722017-03-16T11:17:25.698-04:002017-03-16T11:17:25.698-04:00Actually, there's a proposed patch for that wh...Actually, there's a proposed patch for that which I thought was dead for this release when I wrote the above, but the patch author wrote back some more things about it, so maybe it's not so dead. I have to look at it again. If it doesn't happen in v10, I'm guessing we'll try to make something happen for v11.Robert Haashttps://www.blogger.com/profile/08393677427643988650noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-45882326107790302272017-03-16T09:31:36.908-04:002017-03-16T09:31:36.908-04:00Nice work. Are there plans to support parallel que...Nice work. Are there plans to support parallel query with Inserts. For example create tables as select where the select is run in parallel. Anonymoushttps://www.blogger.com/profile/05695570742457020064noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-6113350709326569832017-03-15T21:15:09.178-04:002017-03-15T21:15:09.178-04:00It's certainly true that every time we add com...It's certainly true that every time we add complexity to the executor, the plans get more complicated. Worse, every new possible plan type we add increases the chances that the query planner will get it wrong, because the total number of options has increased and there's still only one of them that is the fastest. On top of that, the way we're doing planning right now is fairly basic, and no doubt could stand to be improved.<br /><br />Despite all that, I think we're fairly clearly going in the right direction. The new plan types are fairly straightforward generalizations of things we already have, and I think people will get used to them fairly quickly. Even if they don't, the only way to avoid adding the complexity is to skip adding the features, and I don't think anybody wants that.Robert Haashttps://www.blogger.com/profile/08393677427643988650noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-41162083728165648662017-03-15T07:53:26.276-04:002017-03-15T07:53:26.276-04:00Thanks for this read. Very inspiring. Kudos, Enter...Thanks for this read. Very inspiring. Kudos, EnterpriseDB & PostgreSQL - among the most substantial open-source projects.stelfhttps://www.blogger.com/profile/12225084537620761819noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-8746067102387096212017-03-15T07:17:23.720-04:002017-03-15T07:17:23.720-04:00Looking forward to all this, but I fear that under...Looking forward to all this, but I fear that understanding explain statements and predicting a query's performance profile has become much more complicated ? Hopefully only humans will get confused, and the query planner will do a good job.vdphttps://www.blogger.com/profile/07836622731674773883noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-11170359380741720522017-03-15T06:03:00.567-04:002017-03-15T06:03:00.567-04:00Nice write up Robert!
Parallel create index conc...Nice write up Robert! <br /><br />Parallel create index concurrently sure would have been useful to me last week when I dropped an index that was unexpectedly still in use, and mostly took down Bitbucket for the 15 minutes it took to rebuild it. (facepalm)Anonymoushttps://www.blogger.com/profile/02278780378574470222noreply@blogger.comtag:blogger.com,1999:blog-20038672.post-68347388692030371312017-03-14T15:38:26.997-04:002017-03-14T15:38:26.997-04:00Interesting read, thank you.Interesting read, thank you.Anonymoushttps://www.blogger.com/profile/11550598468236049015noreply@blogger.com