Question regarding Solaris LVM state replica databases
From: Paul Boven (p.boven_at_chello.nl)
Date: 04/29/03
- Previous message: ahaukin_at_hushmail.com: "Followup: jumpstart fails at sysiddns when using LDAP"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: Tue, 29 Apr 2003 13:39:40 +0200 To: sunmanagers@sunmanagers.org
L.S.,
I'm installing an E450 which will have 20 disks, 12 filesystems. We will
use DiskSuite aka Solaris LVM for mirroring: each disk will be mirrored
onto another disk on another controller.
I'm trying to come up with the correct number of metadata-replicas I
would need for this. The documentation has this to say about it:
> Too few state database replicas relative to the number of mirrors
> might cause replica I/O to impact RAID 1 volume performance. That is,
> if you have a large number of mirrors, make sure that you have a total
> of at least two state database replicas per RAID 1 volume, up to the
> maximum of 50 replicas per disk set.
This system will have 12 Raid-1 volumes (some will be a simple mirrored
partition, others will mirror a 4-disk stripe), so I'd need 24 metadb's?
However, this seems a bit excessive to me, to have the state database
replicated over so many different disks.
The documentation also states that:
> If you have a RAID 1 volume that will be used for small-sized random
> I/O (as in for a database), be sure that you have at least two extra
> replicas per RAID 1 volume on slices (and preferably disks and
> controllers)that are unconnected to the RAID 1 volume for best
> performance.
Of course, the machine in question will be a database server, and 4 of
these RAID-1's will be doing heavy IO. So this would add another 8
replica's? That would get me up to 32.
I could for instance set this up so every disk has 2 replica's which
gets me 40 of them. Another setup might be to use 1 replica per drive,
and put 3 additional replica's on each of the drives that will have less
IO. That, however, makes those 4 drives a lot more 'important' in the
majority consensus algorithm in case I have diskfailure(s) in the
machine. And I'm not too sure that having 4 metadb replicas in slice 7
would be a good idea.
What would your reccommendation be? And why?
Regards, Paul Boven.
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
- Previous message: ahaukin_at_hushmail.com: "Followup: jumpstart fails at sysiddns when using LDAP"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
|
