Do 10Gbit iSCSI and 8Gbit FC change how you think about your strategy for managing your data in the future?
10345 pts.
0
Q:
Do 10Gbit iSCSI and 8Gbit FC change how you think about your strategy for managing your data in the future?
Do 10Gbit iSCSI and 8Gbit FC change how you think about this and your strategy for managing your data in the future?
ASKED: Sep 3 2008  5:31 PM GMT
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
0
10345 pts.
0
A:
 RATE THIS ANSWER
+4
Click to Vote:
  •   4
  •  0
  • AddThis Social Bookmark Button
Well, since both these protocols are really specific to disk, and current 4Gb FC is barely 10% saturated by the heaviest workload I can generate, no.

I suppose for anyone doing massive sequential throughput to disk might be interested, but I haven't worked on any of those applications.

You can add your thoughts below.
Last Answered: Sep 18 2008  9:17 PM GMT by ITKE   10345 pts.
Latest Contributors: OpenSystemsStorageGuy   545 pts.
0
0
Discuss This Answer:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _



_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

JimmyIT   1260 pts.  |   Sep 8 2008  2:00PM GMT

I agree with the opensystemstorageguy . For my somewhat small company i haven’t changed any strategy yet. I am sure something new will be out in a few years.

 

Eric42   95 pts.  |   Sep 10 2008  8:47PM GMT

Many users find themselves in your position: FC4 is working well for them, their FC4 pipes still have alot of headroom, and today’s price premium for FC8 doesn’t make sense at this point in time. As opensystemstorageguy pointed out, unless you have a lot of sequential IO (or just many truly large transactional applications), there may not be a compelling reason to pay the price for FC8 at this point in time.

Currently, the industry is in the middle of the FC8 transition. And, where the FC2-to-FC4 transition was practically cost point replacement, the optics on FC8 are still at a premium above FC4. Also, most storage arrays do not yet support FC8 or 10Gb iSCSI targets, so many customers don’t see the need to deploy FC8 or 10Gb iSCSI infrastructure just yet.

One area where users can find a benefit in deploying FC8 or 10Gb iSCSI is in virtualized environments, where a large number of large VMs are running on a single physical server. The larger pipes of FC8 and 10Gb iSCSI can help get the IO off the server. Switches can then direct IO to different storage arrays.

-eric @ dell

 

OpenSystemsStorageGuy   545 pts.  |   Sep 11 2008  4:30PM GMT

Actually, the bottleneck in a virtualized environment’s storage stack is more likely to be the physical disk, not the bandwidth between the virtual workloads and the disks.

One application that obviously benefits from this throughput I forgot to mention is backups- tape libraries and vtl are both capable of eating up a large amount of an 8Gb FC pipe.

 

DaveGraham   230 pts.  |   Nov 6 2008  7:14PM GMT

I’ll take it a step further and say that based on my work with customers, 2Gb FC links are just now starting to become saturated. GigE, however, is something that would absolutely benefit from the change to 10Gbe. Get rid of latency issues and really poor TCP/IP overhead issues and you’ve got a very decent pipe that can maintain your infrastructure (Core) for quite a few years.

cheers,

Dave Graham

 

DaveGraham   230 pts.  |   Nov 6 2008  7:16PM GMT

@Eric42,

the price premium of 8Gb over 4Gb is about 10%…that’s quite different than the jump between GigE and 10Gbe (about 25-30%).

cheers,

Dave Graham

 
0