[DRBD-user] difference in speed when using external metadata

Dave Graham daveg at cirtas.com
Fri Jan 14 14:11:26 CET 2011

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Your supermicro controllers are bog-stock LSI SAS controllers so,
functionality will be derived from their driver tree.  Supermicro's UIO
cards are essentially inverted PCIe cards so, no FUNCTIONAL changes, just



Dave Graham
Senior Systems Engineer
Cirtas Systems
75 E Santa Clara Street, Suite 950
San Jose, CA 95113
(978) 239-2489
Dave at cirtas.com    
Follow the latest at the Cirtas Blog <http://www.cirtas.com/blog>!
 <http://twitter.com/cirtas> <http://www.facebook.com/Cirtas>

On 1/13/11 3:07 PM, "Mark Morlino" <mrmorlino at alaska.edu> wrote:

>On Mon, Jan 10, 2011 at 10:31 AM, Bart Coninckx
><bart.coninckx at telenet.be> wrote:
>> On Monday 10 January 2011 19:40:17 J. Ryan Earl wrote:
>>> On Sun, Jan 9, 2011 at 10:03 AM, Bart Coninckx
>> <bart.coninckx at telenet.be>wrote:
>>> > Hi all,
>>> >
>>> > am reading the Novell docs on DRBD and one of the ways they mention
>>> > speed
>>> > up DRBD is using external metadata. Possibly a hard question to
>>> > but could someone indicate by what degree this would enhance speed
>>> > (provided using
>>> > a sufficient fast disk for the metadata, like a SSD)?
>>> >
>>> > thx!
>>> >
>>> > Bart
>>> A RAID controller with a backed-write cache essentially nullifies
>>> any advantage there.  A RAID controller that can instantly return on
>>> FUA-writes is what you want.
>>> -JR
>> Hi,
>> so the essence of your answer is that this is not a matter of having
>> internal or external, but rather of having the right RAID controller?
>> Do you have examples of these in the major brands (HP, Dell, ...) ?
>> thx!
>> B.
>> _______________________________________________
>> drbd-user mailing list
>> drbd-user at lists.linbit.com
>> http://lists.linbit.com/mailman/listinfo/drbd-user
>I would be interested in this information too (in my case I'd be
>interested in any supermicro controllers that meet this requirement).
>If I understand correctly, an FUA-write is a write-through request.
>Doesn't that mean that the controller should not return until
>everything has been physically written to the disk? It seems to me
>that in order to return instantly on an FUA-write, the controller
>would have to ignore the FUA request and return once the data has been
>written to the cache. I don't see how it can both force physical disk
>access and return instantly. Perhaps I'm missing something, it
>certainly wouldn't be the first time.
>drbd-user mailing list
>drbd-user at lists.linbit.com

More information about the drbd-user mailing list