Brocade has 2-year Fibre Channel headstart on rival Brocade is making 16gig Fibre Channel hay while Cisco is still lagging up to two years behind, having made a wrong bet with FCoE. Brocade's chief technology officer, Dave Stevens, presented at an investors' meeting attended by Aaron Rakers of Stifel Nicolaus, who reports that Stevens: "highlighted a faster-than-typical …
Cisco's "mis-step"? I know a lot of companies (I work for one) that are investing in FCoE gear -- specifically, the Nexus 5000/7000 hardware, along with FCoE adapters in the VMware farms.
Where are the studies that show market adoption rates between FCoE and 16Gbps? Who the hell is Aaron Rakers and why should I care what he says?
This seems like a plug for a company that is struggling financially and looking for a buyer *cough* Brocade *cough*
According to IDC FCoE adoption will be about 1% in two years time.
Unless your goal is to be a Cisco only shop there are no reasons for switching to an expensive, complex and under performing technology. A lot of people who can operate a calculator knows that.
One selects technology for a 5 year time frame, the technology that will certainly exist in 5 years time is FC.
FCoE not so sure.
I'm not sure why he thinks Cisco is the only competitor to Brocade. The big FC vendors have been Qlogic, Brocade, and McData. With the later being bought by Brocade. I know you can buy Cisco FC kit, but have yet to see it in the wild.
Qlogic claims they'll have their CNA switch that supports 16Gb out mid next year which by the authors math puts them about 1.5 years ahead of Cisco. Not sure why the article completely forgot about Qlogic. Usually the comparisons are between Brocade and Qlogic.
FCoE was trapped at top of rack until the multi-hop additions were added - it's still very limited compared to FC, so I'd fully expect FC to reign for a while longer in the core of large SANs. Not forever though, maintaining a whole separate SAN infrastructure just doesn't add up with 10 and soon 40 gig Ethernet cheaper and with QoS/lossless extensions, Brocade better stockpile that hay...
Surely by the time 16Gb FC arrives, 40Gb ethernet will be prevalent (or at least similar cost)?
Why are people still obsessed with FC when ethernet speeds are/will easily outpace FC?
Admittedly iSCSI is not the best storage protocol, but some of the 10Gb enhancements might make iSCSI more acceptable, and there's always ATA Over Ethernet (Coraid seem to be the main vendor at the moment) which from what I understand is low latency layer 2 transport.
As for FCoE - why not just drop the "FCo" part and just go for the "E"?
On what basis do you assume that future 40Gb E will be on par with 16 Gb FC cost wise?
YMMV, but we did some cost calculations and a FCoE capable Nexus 5000 came out with 3 times the port cost compared to a Brocade 5000 8Gb switch.
Port-to-port latency on a Brocade is 700 ns.
From what I have been able to find out port-to-port latency on a NX5000 is 3.2 microseconds. That is 4.5 times as much. And this is before the added overhead of decapsulating the FC frame and doing a zone lookup.
The only cost savings I regard as valid with FCoE is that you save the $50 and the 5 minutes of work to stretch another fibre between edge and core.
The cost benefit also does not factor in that the majority of folks that move their data on to the network segregate it for performance -- usually by an additional port/vlan. As well as additional port for redundancy, thus not reducing the port/cable count. And if we compare the cost of the Cisco CNA kit to our current Qlogic/Brocade SAN infrastructure we have yet to see any kind of cost benefit to removing our SAN infrastructure. As well as the fact that we'd take a performance hit due to the latencies and overhead. FC may be replaced down the road, but I don't see it to the network. If anything replaces FC, it'll be infiniband.
By running FCoE, it's not as efficient. Packet encapsulation and a higher latency. Usually you want your IO fast as well as with low a low latency.
The future will hold a converged infrastructure. But it will be fast and have low latency. Hello Infiniband! It's already making it's way into more and more non-HPC datacenters. You just don't hear as much about it as it's not promoted heavily by Cisco.
16Gbps FC is already here - been shipping a couple of months as in the article (18% of Brocades FC shipments)
When will 40Gbps be here realistically and widely adopted?
10Gbp adoption is still pretty slow as it costs more than 8Gbps FC purely from an aquisition cost. I appreciate you can get TCO savings through convergence though.
Don't think FC is dead quite yet and will remain as a transport mechanism for datacentre storage for a fair while yet.
Cisco have always been behond the curve on FC SAN performance - software wise they are pretty good though
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
