[CentOS] strange partitioning problem

Boris Epstein borepstein at gmail.com
Wed Apr 25 02:20:00 UTC 2012


On Tue, Apr 24, 2012 at 10:17 PM, John R Pierce <pierce at hogranch.com> wrote:

> On 04/24/12 2:24 PM, Eero Volotinen wrote:
> > Anyway, with 18TB of raid 6 might cause performance problems .. raid
> > 60 or raid 10 looks much better solution.
>
> 18TB of raid6 is only 8 3TB drives (6 + 2 parity).... thats not an
> unreasonable size for a raid6 set.
>
> for the 81TB nearline storage boxes I built for a data archive
> application, I formatted each box (using a megaraid sas2 9260-8i card
> and 36 3TB SAS drives) as 3 x 11 x raid60, plus 3 hot spares, and then
> made one big XFS on that.   Its working out quite nicely.    I think 10
> or 12 disks per raid5 or 6 is a good upper limit.    rebuild time from a
> single failure was 12 hours, and from a double failure in the same raid
> plex, 18 hours.
>
>
>
> --
> john r pierce                            N 37, W 122
> santa cruz ca                         mid-left coast
>
>
 Actually in our case we've got 12 2TB drives - but still I think John's
point stands.

Boris.



More information about the CentOS mailing list