vinum in 4.x poor performer?

Mark A. Garcia mag at hamletinc.com
Wed Feb 9 10:38:49 PST 2005


Marc G. Fournier wrote:

>
> Self-followup .. the server config is as follows ... did I do maybe 
> mis-configure the array?
>
> # Vinum configuration of neptune.hub.org, saved at Wed Feb  9 00:13:52 
> 2005
> drive d0 device /dev/da1s1a
> drive d1 device /dev/da2s1a
> drive d2 device /dev/da3s1a
> drive d3 device /dev/da4s1a
> volume vm
> plex name vm.p0 org raid5 1024s vol vm sd name vm.p0.s0 drive d0 plex 
> vm.p0 len 142314496s driveoffset 265s plexoffset 0s
> sd name vm.p0.s1 drive d1 plex vm.p0 len 142314496s driveoffset 265s 
> plexoffset 1024s
> sd name vm.p0.s2 drive d2 plex vm.p0 len 142314496s driveoffset 265s 
> plexoffset 2048s
> sd name vm.p0.s3 drive d3 plex vm.p0 len 142314496s driveoffset 265s 
> plexoffset 3072s
>
> bassed on an initial config file that looks like:
>
> neptune# cat /root/raid5
> drive d0 device /dev/da1s1a
> drive d1 device /dev/da2s1a
> drive d2 device /dev/da3s1a
> drive d3 device /dev/da4s1a
> volume vm
>  plex org raid5 512k
>  sd length 0 drive d0
>  sd length 0 drive d1
>  sd length 0 drive d2
>  sd length 0 drive d3
>
It's worth pointing out that your performance on the raid-5 can change 
for the better if you avoid having the stripe size be a power of 2.  
This is especially true if the (n)umber of disks are a 2^n.

Cheers,
-.mag


More information about the freebsd-questions mailing list