[clug-talk] WD Green HDD's with ZFS
gustin at meganerd.ca
Sun Mar 25 21:32:28 PDT 2012
On Sun, Mar 25, 2012 at 8:00 PM, Royce Souther <osgnuru at gmail.com> wrote:
> I have a few file servers that I am using Western Digiatl Green HDD's with
> LVM. I am not using RAID1 because that would burn out a Green drive very
> quickly. My servers are identical clones of each other and I like that
> better then RAID1 because if a power supply dies the other server keeps the
> work going until I can get the dead one fixed.
> I like the idea of redundancy in the actual servers, but I have never
heard of a RAID1 killing a WD Green drive. I have a lot of these in service
(both at work and at home) in RAID5 and RAID10 configurations and they
simply hum along. If anything the reduced heat of these drives should
result in extended lifetimes.
I one big issue with LVM is having to unmount the file system to resize the
> partition. Every time I add a drive and grow the system it takes longer and
> longer to go through the steps of fsck'ing the partiion and then resizing
> it. It takes most of a day now to do this. ZFS does not have this problem.
> As I understand it ZFS is able to resize the file system on the fly.
> Mirrored Vdev would be like using RAID1 and that would kill a Green drive
> but just using a ZFS to have the abilty to grow the FS on the file would be
> good enough and should not hurt the Green drives. Does anyone know if using
> ZFS not using mirror would be safe to use on drives like these?
> First, LVM is not a file system and can be resized without unmounting
since a Logical Volume is never mounted directly. For that you need to
layer a file system on top. Very few of the Linux file systems support
resizing on the fly. To my knowledge only XFS supports this. At home I no
longer really use LVM though this may change since the only feature that I
really like is copy on write. For grouping disks together (for RAID or
JBOD) I just use mdadm.
I have only messed with ZFS a little. It was a PITA to get and then keep
running (the joys of non-GPL friendly code and kernel upgrades). Unless it
gets released under a GPL friendly licence I suspect that it will stay in
the Oracle and Apple worlds. I am really not a fan of ZFS unless you have
tiers of storage. I like that it can transparently (to the user) move data
between really fast (RAM), pretty fast (SSD), and then spinning metal. It
is a neat idea though I am sceptical that it is ready for prime time. The
ZFS team does have a lead on BTRFS but I would stick to whatever can get
included into the mainstream kernel. It will cause far fewer headaches in
the long run.
> Easy, fast GUI development.
> clug-talk mailing list
> clug-talk at clug.ca
> Mailing List Guidelines (http://clug.ca/ml_guidelines.php)
> **Please remove these lines when replying
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the clug-talk