Comments on: FCoE, Will you believe the hype?

By: Jesse

Jesse — Sun, 10 Aug 2008 16:54:45 +0000

Very interesting. It’s been a while since I’ve had the time to post to my own blog, let alone browse the others. So far I’ve found very little real-world implementations of FCoE. Seems to me that that particular implementation would be inefficient as you’re wrapping a SCSI packet in an FC packet, then turning right around and wrapping that up in an ethernet packet. Seems like the overhead would be excessive.So far I’ve found iSCSI to be useful in small implementations. Much easier and less expensive to put a dedicated $1,000 Ethernet switch (a good one, you don’t want to use a crappy switch for storage traffic) than a $10,000 fibrechannel switch.

By: Deepak Munjal

Deepak Munjal — Tue, 06 May 2008 17:37:32 +0000

I've read some valid points but let me expand on some of the other ones. The fact is that customers are looking for a converged solution. Maintaining multiple parallel networks has never made sense. You only have to look at networks in the past that have been converged (FDDI, ATM, Token Ring, X.25, etc.) to understand that the future will be a single network to attach all types of devices. Historically, IP and Ethernet have seen the most success here. No one is advocating one converged solution over another. FCoE makes sense for those who have invested millions in Fibre Channel and provides an elegant migration solution. iSCSI appeals to many small and medium businesses that want to build a SAN but don't want to make an investment in Fibre Channel. Cisco is investing in all of these solutions to give customers the choice. The MDS 9000 will continue to be our flagship Fibre Channel solution and will support FCoE. The Nexus platforms support Data Center Ethernet which provide the platform to converge storage and data traffic whether you're using iSCSI or FCoE.

By: Etherealmind

Etherealmind — Tue, 06 May 2008 06:28:23 +0000

I agree with most of what you say, but I also think that FCoE looks a 'transition technology' for all those people who have invested personal energy into the FC. I have seen quite a bit of reaction from people who have FC skills and can see that disappearing and they will lose influence in the their organisations. Those folks are not going to give up easy and will point the finger at the Networking Team and say that IP networking is not ready. I have put some design articles for iSCSI on my blog. Maybe you will find them interesting. Etherealmind

By: Nigel

Nigel — Fri, 02 May 2008 19:19:48 +0000

Its not that long ago that most of us only connected our high performance servers to the SAN. This connectivity was was over dedicated fibre, running a purpose written protocol to dedicated high speed (and cost) switches into subsystems stuffed with high speed FC disks running RAID1/10…….

What will we have tomorrow…… Nearly all of our server estate crammed onto the ~~SAN~~ shared network via iSCSI and FCOE over a shared medium, through shared switches (share processor clock cycles etc with TCP/IP traffic) and into subsystems loaded with everything from SSD to FC to SATA running every RAID level you could care to imagine and a few you’d rather not imagine.

At least you will get predictable performance……. Predictably bad!

Actually, may be it wont work like that….. I haven’t done enough research into it yet, but it certainly seems to be the way the industry is going. While Ive always liked EMC’s pitch on SSD, I also see that it brings its own problems. The industry seems to be desperate to cram as many products, protocols and features as possible into a single entity, be it piece of tin, apiece of cable, a switch or array…….. all under the name of consolidation. The thing is, I personally prefer expansion and widening my horizons rather than cutting back and consolidating. But the market dictates.

BTW I don’t see the big banks and the likes being keen on FCOE. Too risky in my opinion. May be 10 years after the rest have adopted it!?

By: billy bathgates

billy bathgates — Tue, 29 Apr 2008 23:20:58 +0000

The comment about cisco is disconcerting. Cisco keeps doing their damndest trying to make storage management and storage groups be seen as part of 'networking' in corporate organizational charts. Gee wonder why they would want to do that? I'm pretty sure I don't wnat cisco deciding what the future of storage networking will be (not that I have any control over the matter, I can still bitch though:-)

By: billy bathgates

billy bathgates — Tue, 29 Apr 2008 23:14:41 +0000

Amen Chris, In my shop, it is quite rare that a 4G link is more than 50% utilized, although lets face it, it is nice to be able to use fewer ISLs with the higher bandwidth (and by "higner bandwidth, I mean 4G). I bet the number of shops that would really benefit from 8G is small, and even there, the number of 8G links they need is a small percentage of their SAN.

By: Chris M Evans

Chris M Evans — Tue, 29 Apr 2008 20:29:49 +0000

I think customers have more fundamental issues to solve rather than whether they should use host based 8GB or FCoE. Perhaps people out there should take a step back and look at solving all the existing issues in their environment first, rather than putting in the new toys!

By: Barry Whyte

Barry Whyte — Fri, 25 Apr 2008 21:22:16 +0000

I guess the major question is if you can use the same Ethernet infrastructure for your TCP/IP connectivity and your FCoE connectivity. A la DCE. However, unless there is a major benefit (and the major issues with CPU overhead are overcome that have dogged iSCSI’s break through in the enterprise market – amazing that even offload iSCSI cards save very little of that overhead) then FC as we know it will be here for a while longer.

8Gbit is here and now, but who is asking for it?