All this talk...
... about fibres, fibre channels and fabric puts me in mind of two cups and a piece of string. It's a grossly unfair idea but... it's just stuck, I can't help it.
Here's a thought: Fibre Channel has begun its death march, with physical fabrics under notice from FCoE, and FC-interface hard drives under notice from SAS. You might not agree, but here's the argument in favour: Internal array Fibre Channel Currently Fibre Channel (FC) has four main incarnations, hardware and software inside …
With modern crap-load-'o-cores CPUs, I predict that you will see the end of HBAs. Once you have a lot of CPU power that can be easily divided, offloading the I/O stack to a core instead of a dedicated (read: expensive) chip seems like a winner idea in my book.
Certainly this will take a while, as a complete SCSI stack is pretty darn complex, and I would not trust the stability of such a stack for a couple of years. (Similar to the early days of iSCSI.)
Intel didn't write an open-source software implementation of FCoE because they are merely a bunch of charitable folks.
If you want another prediction: After FCoE displaces native FC in the SAN and gets decision makers comfortable with Storage over Ethernet, it will be quickly followed by a transition to iSCSI, which is not quite so picky about network quality. It is also much easier to find an IP admin instead of a SAN admin.
"I reckon the odds on physical FC death are one in four with FCoE death being much less likely, say one in 20. What do you think?"
FCoE is just a thrash of death of vendors of expensive and inconvenient technology. FCoE doesn't really have any advantages over ATAoE, and ATAoE is cheaper and more mature. iSCSI has a big advantage over both in that it can be routed via IP, because it operates on a higher level on a networking stack.
FCoE's death being much less likely? FCoE is destined by design to be stillborn by design!
This post has been deleted by its author
Anyone remember Network Systems Hyperchannel and RFCs 1044 and 1223? Interesting - actually I would have thought the speed going up even more and with four trunks but they got bought out.. Light will be faster (photons versus electrons) but maybe we have to wait - again? NSC HC was actually more than just disks (or any I/O) - and very easy to write drivers / applications. Of course, today those skills are (maybe) only on vendors, no user / customer has any more skills to create that level systems, they buy and operate (sorry, administer) what they are sold - sad!
At one time I had a "smart" terminal with 6 3270 terminal sessions. Then the PC showed up and replaced the "converged" smart terminal into a paperweight. Every time massive convergence to a single connectivity option occurs, a new option is created on new technology. I already made the investment in FC hardware.<br><br> If what you say is true, then supporting that infrastructure will only get cheaper over the next 5 years. Then I will look at the new connectivity options.