tag:blogger.com,1999:blog-9053536529694784563.post3739856605128489833..comments2020-05-07T00:33:57.272-07:00Comments on Database Explorer: Hot StandbySimon Riggshttp://www.blogger.com/profile/06017750505968534813noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-9053536529694784563.post-28661576084297173032009-09-09T15:16:45.036-07:002009-09-09T15:16:45.036-07:00Another feature I am eager to see is read-only tab...Another feature I am eager to see is read-only table/tablespace. Right now in my production database there are about 90% of data which are read-only. From time to time, autovacuum picks up these table to prevent XID wrap-around. If read-only table/tablespace is available, I hope I can put these tables in a filesystem that can shared among several database nodes. It will save a lot of disk space.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9053536529694784563.post-36711244512971179492009-09-08T10:12:14.929-07:002009-09-08T10:12:14.929-07:00Thanks for your comments, anon. I guess if it hurt...Thanks for your comments, anon. I guess if it hurts, you're probably doing it wrong. It's true that there is no "switchback" command and that you have to re-issue the commands as if you were starting from scratch. The underlying technology is for you to choose, so many people choose tools like rsync which can significantly reduce the time to re-run the switchback operation by avoiding transferring data that is already the same. That's why most people think what you've described isn't a problem at all and that leaves Hot Standby as a popular feature. I'm not sure why some people overlook this, but they do, so you're not alone.Simon Riggshttps://www.blogger.com/profile/06017750505968534813noreply@blogger.comtag:blogger.com,1999:blog-9053536529694784563.post-15259321425993548182009-09-08T09:47:39.522-07:002009-09-08T09:47:39.522-07:00One of the current issues with creating a Warm-Sta...One of the current issues with creating a Warm-Standby within PostgreSQL for a disaster recovery configuration, is the lack of the capability to easily switch back and forth from the master (primary) to the slave (alternate). This situation is desirable for several uses: disaster drills, network patching, OS patching, hardware updates, … If such a capability existed, then transactions against the primary would be stopped, the roles of primary and alternate are temporarily reversed (slave promotion), and the transactions are restarted against the new primary (previous alternate or slave). Then the roles can be reversed back. <br /><br />As this is not currently available within PostgreSQL, alternative (inferior replication) DR configurations have to be evaluated.<br /><br />My feeling is the improving of the Warm-Standby is more important than the creation of a Hot-Standby. Making PostgreSQL's Warm-Standby more akin to Oracle's Data Guard. Yes, Oracle also released a Hot-Standby configuration within its Data Guard technology referred to as Active Data Guard (with an additional cost).Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9053536529694784563.post-25068789482048442942009-09-07T04:37:08.444-07:002009-09-07T04:37:08.444-07:00hahaha, just works is a problem? i like those sort...hahaha, just works is a problem? i like those sort of problems!!! bring it on, would love to see this in action with postgresUnknownhttps://www.blogger.com/profile/15104599982741309220noreply@blogger.com