32 Gig Fibre Channel takes another step towards reality The Fibre Channel Industry Association (FCIA) has announced that a 32 gigabits per second version of the standard is just a little bit closer to reality. The Association has let it be known that the “INCITS T11 standards committee has recently completed the Fibre Channel Physical Interface - sixth generation (FC-PI-6) industry …
I'm surprised that FC hasn't aligned itself more with Ethernet. Ethernet has 40Gbps and 100Gbps with 400Gbps coming soon. By doing so, faster FC speeds would become available far sooner. According to the chart, 256Gbps FC should be finalized by 2022. 400Gbps Ethernet will be available before that.
You're wondering why an industry association isn't showing a roadmap using the physical layer of its mortal enemy? The closer FC comes to ethernet, the more likely people are willing to accept the limitations of ethernet in exchange for the speed, and FC goes away. Where is the FCIA if that happens?
The main reason that FCoE hasn't gained much traction in my experience (as a consultant who often works in the storage field) is that the group managing the network and the group managing storage (whether or not there's a dedicated group or there are part of a larger server-related group) are different.
Add that to the fact that no enterprise will trust running their SAN on the same network as, you know, network data (except possibly storage related data, like a dedicated backup network) so you end up having separate infrastructure anyway, and see why FC looks to be around for a while longer, even as ethernet speeds past it. Ethernet will take over as a cheap low end SAN, and slowly work its way up as people gain trust in it, but it'll take quite some time before it takes over more conservative enterprise SANs.
Given that vBlock uses FC extensively. Even NetApp's FlexPod has an FC option that is frequently deployed. The need for block storage over SCSI isn't going to disappear any time soon and FC remains the best blend of cheap/reliable/fast on the market once you get outside of disk in the server itself. iSCSI can be cheaper, but at the loss of speed and reliability. iSCSI can be faster and just as reliable, but then it costs more than FC.
"LOL, you don´t work for a company selling fibre channel switches with licensed ports and carepacks for the switch and the licensed ports, don´t you?"
Nice double negative, but I get your drift. No, I work for a company that designs and deploys both iSCSI and FC networks and storage. I challenge you to show my statement is incorrect. Find an Ethernet switch cheaper than an FC switch on a per-port basis. It has to be 10Gb or it won't be faster than even 8Gb FC (let alone 16Gb). And, it will have to support DCB/CEE or it won't be as reliable as native FC. Don't forget the optics in the pricing or it is not apples-to-apples. Good luck, you will need it.
Price comparison is always a bad idea, isn´t it? So let´s look in one of my main webshops:
1x HP 5900AF-48XGT-4QSFP+ (JG336A) with 48x 10GBase-T, 4x 40GbE QSFP+: 5821,-€
1x HP HP X711 Front to Back Fan Tray(JG552A): 102,-€
2x HP 650W AC power supply (JC680A): 438,-€
1x HP carepack U6P29E: 5 years, next business day exchange, software 2 hours: 2298,-€
of course without VAT
Update:
sorry, how could I forget "the optics" ;-)
- 4x HP X140 40G QSFP+ SR4 (JG325A): 3750,-€
Cat6 for a a5900. A a5900 is a core switch. Were you normally connect top of rack switches to and bladeenclosures and virtual connect 10gb ethernet if it must be. Unless you want to lay 15meter cat6 cables to this switch i will choose fibercables any day of the week, because then i can put them in the same goat? As my san fiberchannelcables instead of cumbersome, because to tick, cat6.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
