Re: Does anyone use in ram postgres database?
От | Alan McKay |
---|---|
Тема | Re: Does anyone use in ram postgres database? |
Дата | |
Msg-id | 844129e81003260706q3d1ce1fat811921732f720d97@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Does anyone use in ram postgres database? (Scott Marlowe <scott.marlowe@gmail.com>) |
Ответы |
Re: Does anyone use in ram postgres database?
Re: Does anyone use in ram postgres database? Re: Does anyone use in ram postgres database? |
Список | pgsql-general |
On Thu, Mar 25, 2010 at 4:15 PM, Scott Marlowe <scott.marlowe@gmail.com> wrote: > These questions always get the first question back, what are you > trying to accomplish? Different objectives will have different > answers. We have a real-time application that processes data as it comes in. Doing some simple math tells us that a disk-based DB cannot possible perform fast enough to allow us to process the data. > Now, if your pg_xlog directory is a problem, then you either need > bigger faster hard drives, or your data is more transient in nature > and you can recreate it and you put the whole db into RAM. When we only saw a 3x improvement in speed with the RAM based DB, we were still seeing a fair bit of disk activity but were not sure what was going on. Then we thought about pg_xlog and moved it to RAM as well, but as I recall still not a great improvement. We are trying a test right now where "initdb" was run against /ramdisk/data so that absolutely everything should be in there. Will report back with results. We are also about to try another test with a regular disk-based DB and fsync turned OFF > Note that the query planner wasn't designed with RAM as the storage > space for pg, so it might make some bad decisions until you adjust > postgresql.conf to stop that. and then it still might make some bad > decisions. What thinks might need adjusting? thanks, -Alan -- “Don't eat anything you've ever seen advertised on TV” - Michael Pollan, author of "In Defense of Food"
В списке pgsql-general по дате отправления: