Re: swapcache-style cache?

----- Цитат от Andrea Suisani (sickpig@opinioni.net), на 28.02.2012 в 09:54 ----- <br /><br />> On 02/28/2012 04:52
AM,Rob Wultsch wrote: <br />>> On Wed, Feb 22, 2012 at 2:31 PM, james wrote: <br />>>> Has anyone
consideredmanaging a system like the DragonFLY swapcache for a <br />>>> DBMS like PostgreSQL? <br
/>>>><br />>> <br />>> https://www.facebook.com/note.php?note_id=388112370932 <br />>> <br
/>><br />> in the same vein: <br />> <br />> http://bcache.evilpiepirate.org/ <br />> <br />> from
themain page: <br />> <br />> "Bcache is a patch for the Linux kernel to use SSDs to cache other block devices.
It'sanalogous to L2Arc for ZFS, <br />> but Bcache also does writeback caching, and it's filesystem agnostic. It's
designedto be switched on with a minimum <br />> of effort, and to work well without configuration on any setup. By
defaultit won't cache sequential IO, just the random <br />> reads and writes that SSDs excel at. It's meant to be
suitablefor desktops, servers, high end storage arrays, and perhaps <br />> even embedded." <br />> <br />> it
wassubmitted to linux kernel mailing list a bunch of time, the last one: <br />> <br />>
https://lkml.org/lkml/2011/9/10/13<br />> <br />> <br />> Andrea <br />> <br /><br />I am successfully
usingfacebook's flashchache in write-through mode - so it speeds only reads. I have seen 3 times <br />increase on TPS
fordatabases that do not fit in RAM. I am using Intel X-25E over RAID10 of 4 SAS disks. I have tested also <br
/>writebackmode but the gain is not so huge and there is a considerable risk for loosing all your data if/when the SSD
fails.<br /><br />Best regards <br /><br /><br />-- <br />Luben Karavelov