LVM + RAID follow up

Derek Atkins warlord at MIT.EDU
Wed Nov 8 17:24:55 EST 2006


Quoting Rich Braun <richb at pioneer.ci.net>:

> "Derek Atkins" <warlord at MIT.EDU> responded to my query:
>>> your swap is *not* configured as RAID?
>>
>> Yeah, that's how I have it configured.  It's much better performance
>> to do it this way (why double the number of writes you have to make when
>> you swap out a page?)..
>
> Hmm...that statement doesn't compute.  The dual writes happen virtually
> simultaneously, last time I did benchmarks the difference between 
> non-RAID vs.
> software RAID1 the write performance difference was something like 2%.
>
> Is there something special about swapping versus filesystem access that would
> cause a bigger performance difference?  Also--I tend to assume that swap
> performance isn't particularly crucial anyway, because if I'm swapping that
> means I need to upgrade system memory.  If overall system performance is
> (virtually) unaffected, then it stands to reason that fault tolerance should
> take priority.
>
> Am I all wet, or do these observations hold true?

I suspect it depends on the types of drives, drive controller, and disk
layout.   I never performed any kind of metrics.   Also, it winds up
cutting your swap space in half (for a 2-disk RAID-1 mirror)..  I consider
swap like scratch space..

> -rich

-derek

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord at MIT.EDU                        PGP key available


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.




More information about the Discuss mailing list