Comments on: Tips and tricks: Memory storage on PostgreSQL http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/ Red Hat Magazine Fri, 8 Aug 2008 20:55:11 +0000 http://lyceum.ibiblio.org/?v=1.0.2 by: Alexander Todorov http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-42174 Sun, 23 Dec 2007 19:37:45 +0000 http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-42174 Peter, Thanks for the heads up. Peter,
Thanks for the heads up.

]]> by: Peter Sinclair http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-41927 Sat, 22 Dec 2007 21:47:46 +0000 http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-41927 Useful tip. I assume the first command "mount -t ramfs none /mount/point" should have mounted at /mount/ramfs as this is the mount point referred to in all later discussion. Useful tip. I assume the first command “mount -t ramfs none /mount/point” should have mounted at /mount/ramfs as this is the mount point referred to in all later discussion.

]]> by: Walt http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-41614 Fri, 21 Dec 2007 23:29:58 +0000 http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-41614 Increasing shared buffers can effectively keep a db in memory. The default configuration has traditionally been way too conservative - am sure someone will eventually get round to writing a script to tune configuration based on a user response to questions and hardware probing. Course if the data is not critical and you just need something lightweight there are application specific options... Increasing shared buffers can effectively keep a db in memory. The default configuration has traditionally been way too conservative - am sure someone will eventually get round to writing a script to tune configuration based on a user response to questions and hardware probing.

Course if the data is not critical and you just need something lightweight there are application specific options…

]]> by: Pavel Stěhule http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-39674 Fri, 14 Dec 2007 10:06:43 +0000 http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-39674 Hello it will works only for selects. Updates and deletes have to be commited to WAL, and if WAL is on normal disk, then you cannot expect any speedup. So. If you want to play with this, you have to put to ramdisk complete database cluster and set fsync = off in postgresql.conf. Else don't use ramdisk and set bigger value of shared_buffers. It has similar efect and its more correct solution. Pavel Hello

it will works only for selects. Updates and deletes have to be commited to WAL, and if WAL is on normal disk, then you cannot expect any speedup.

So. If you want to play with this, you have to put to ramdisk complete database cluster and set fsync = off in postgresql.conf. Else don’t use ramdisk and set bigger value of shared_buffers. It has similar efect and its more correct
solution.

Pavel

]]> by: Alexander Todorov http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-39472 Thu, 13 Dec 2007 09:12:47 +0000 http://www.redhatmagazine.com/2007/12/12/tip-from-an-rhce-memory-storage-on-postgresql/#comment-39472 NOTE: In the example above we dump only database schema. Normally applications processing data in memory don't care about the data itself (because it has a temporary nature) or they are able to re-create the data. If you care about the contents in the database remove the --schema-only parameter in the code above. NOTE: In the example above we dump only database schema. Normally applications processing data in memory don’t care about the data itself (because it has a temporary nature) or they are able to re-create the data.
If you care about the contents in the database remove the –schema-only parameter in the code above.

]]>