himanshu padmanabhi wrote:
> On Wed, Jun 25, 2008 at 8:21 PM, Ross S. W. Walker <
RWalker@...> wrote:
>
> >himanshu padmanabhi wrote:
> >
> >> I am using open-iscsi initiator and IET target.
> >>
> >> I just want to use the exported disk by creating VG's,LV's on
> >> the exported disk from the target.
> >>
> >> Then of what type should I export LUN from the target
> >> side.Fileio or BLockio?
> >
> > Let me ask you these questions:
> >
> > 1. What is the underlying storage type (RAID1, RAID10,
> > RAID5/RAID6, single disk)?
>
> It can be anything out of this.Lv''s will be created
> on anything.Can u specify for all?
>
> > 2. Is it software or hardware RAID?
>
> We are using both.Can u specify for both?
>
> > 3. Does it have a write-back cache? Read-ahead cache?
>
> We used 3ware's hardware RAID.I have to check what it uses?
>
> > 4. What is the application/use on the initiator for
> > this storage (file, database, backup, streaming media)?
>
> Initiator can use it for any above said purposes.
>
> So there isn't any generalised rule for this?
No, if there was then there wouldn't be any configuration needed.
Ok, so here is something that may or may not be helpful, of course as always YMMV:
1) Software RAID (which has no caching) use Type=fileio,IOMode=wb
2) Hardware RAID (with no caching) same as above
3) Hardware RAID (with write-back, no read-ahead) use Type=fileio
4) Hardware RAID (with write-back and read-ahead) use Type=blockio
Exceptions:
1) If you don't have a UPS don't use IOMode=wb
2) If you don't have a rock solid stable single-purpose target, don't use IOMode=wb
3) If you plan on connecting to target over loopback interface, use Type=blockio
4) If you are using the volume primarily for backups or streaming media, which do large sequential IO, use Type=blockio
5) If you are using the volume for databases, which do both random IO and large sequential IO, use Type=blockio
6) If you are using the volume for virtual machines, which do random IO, but request the same data over and over again use Type=fileio
Use LVM on the target to divide storage between the differing types (best practice is to have separate arrays for the differing types) and set the lun types as appropriate.
LVM on the target also has the advantage of being able to do live migrations of storage using pvmove which you will find extremely helpful when it's time to upgrade your arrays and move data over. Also if block level de-duplication and thin provisioning ever get incorporated into LVM then you are set to leverage those technologies as well.
-Ross
______________________________________________________________________
This e-mail, and any attachments thereto, is intended only for use by
the addressee(s) named herein and may contain legally privileged
and/or confidential information. If you are not the intended recipient
of this e-mail, you are hereby notified that any dissemination,
distribution or copying of this e-mail, and any attachments thereto,
is strictly prohibited. If you have received this e-mail in error,
please immediately notify the sender and permanently delete the
original and any copy or printout thereof.
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php_______________________________________________
Iscsitarget-devel mailing list
Iscsitarget-devel@...
https://lists.sourceforge.net/lists/listinfo/iscsitarget-devel
