COMSTAR is development complete and is under test now. The packages  
and source are available on opensolaris and are being used by a  
number of people already and so far we have not had reports of any  
major issues. The current plan is to integrate COMSTAR into nevada  
build 90 which is in the beginning of May this year.

Performance depends upon your reference platform and a number of  
other things. The platform we use internally for performance  
measurement is x4500 (aka Thumper). So far we have been able to  
saturate a 8G FC link with 32k blocksize and CPU utilization of 22%  
doing Reads from the initiator (zfs raidz back end). Same is true for  
Writes. But when running both reads and writes together, you hit the  
PCI-X limit. We are hoping to run a more formal set of performance  
numbers soon and post the results on the project's opensolaris page.

Sumit

On Mar 15, 2008, at 1:26 AM, Emmy wrote:

> Hi all,
>
> I was wondering - is the COMSTAR project considered to be ready for  
> production use?  It looks very interesting and may be the solution  
> some of our issues, but we do not want to risk any data loss.
>
> Also, what are the performance figures expected to be like?  Have  
> any thorough tests been performed in terms of IOPS, data  
> throughput, random r/w tests, etc?
>
> Thanks!
>
> emmy
>
>
> This message posted from opensolaris.org
> _______________________________________________
> storage-discuss mailing list
> [email protected]
> http://mail.opensolaris.org/mailman/listinfo/storage-discuss

_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to