[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
RE: [linux-lvm] Disk striping option
- From: pswartz digitalrom com (Patrick Swartz)
- To: "Andreas Dilger" <adilger turbolinux com>
- Cc: <linux-lvm msede com>
- Subject: RE: [linux-lvm] Disk striping option
- Date: Wed, 8 Nov 2000 14:48:30 -0600
Thanks for responding so quickly. I should add that each JBOD chassis is
attached to it's own ICP card in a Raid 5 + hot spare configuration. This
system will be used in a heavy fileserver environment doing reads-writes
with large files.
>From LVMs point-of-view would it only see "2" drives (the two h/w raid5's)?
Thanks,
Patrick
-----Original Message-----
From: Andreas Dilger [mailto:adilger turbolinux com]
Sent: Wednesday, November 08, 2000 2:28 PM
To: Patrick Swartz
Cc: linux-lvm msede com
Subject: Re: [linux-lvm] Disk striping option
Patrick Swartz writes:
> I have a question about the use of the LVM native striping option. Here
is
> my scenario: I have RH7.0 with an ICP-Vortex Raid controller and 11
Seagate
> 50G hard-drives in a JBOD chassis that is set up as a single VG. An older
> system has the same configuration without LVM and using RH6.2.
> What I want to do is move all of the data from the older system to the LVM
> system and then add the old 11 drives to the LVM system to in affect
double
> my space available.
>
> Also, do you have any recommendation for a File System to use? This is a
> production system so I am a little leery of 0.0.x software, but I am not
> crazy about ext2 fcsk on 1TB of data either.
If this is a production system, I would be very leery of using 11 disks
striped together into a single filesystem. I'd be even more leery of
having 22 disks in this configuration. First of all, you can't get
close to 11x the bandwidth of a single disk, and secondly you will
kill your I/O rate because each of the 11 disks will have to be involved
in every I/O rather than allowing them to each handle separate I/Os.
Also, you run a huge risk of trashing all of your data if a single disk
fails, whereas with the normal concatenated configuration you at least
have some chance to get the data back.
RedHat is running a Beta program with ext3-0.0.3b, and I've been using
it on my systems for several months without problems. Despite the
version number, it is very robust and is only lacking metadata-only
journalling (added in 0.0.5 but it came with a few bugs).
Of course, your choice of filesystem also depends on what you are using
this server for.
Cheers, Andreas
--
Andreas Dilger \ "If a man ate a pound of pasta and a pound of antipasto,
\ would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/ -- Dogbert
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]