FibreChannel dead? Nonsense, says Brocade, here's our Gen 6 kit Brocade's having none of this FibreChannel doomsaying, with the launch of its FibreChannel Gen 6 director family. The launch comprises the Brocade X6 Director, SX6 Extension Blade, and for monitoring, two new enhancements to its Fabric Vision. The X6 Director has 384 ports running at 32 Gbps, and supports aggregation of those …
Gen6 FC appears to have port-to-port latency of 700 nsec, compared to 90 nsec for Omni-Path and 110 nsec for Mellanox Infiniband.
Although Gen6 FC is superficially 700% worse in terms of latency, it is less clear how important that is in terms of solution complexity, service-delivery and total costs. It is not hard to imagine a lot of cases where the maturity of FC combined with 'sub-microsecond latency' is more than 'good enough'.
Its a bit unfair to compare completely different protocols and expect them to be the same.
Infinniband is well known for low latency, short links and high bandwidths and although I've not used Omni-path, this seems to be another HPC technology rather than FC which is a storage technology.
FC by comparison gives reliability; throughput; zone based security and can be extended internationally for storage traffic, so they are really Apples vs Oranges.
FC is a rock solid protocol. Its good to see a new generation arrive. I'd argue that FC is more prevalent than the specialist technologies used in HPC. How many of those HPC protocols exist in the nearest cloud for example ??
FC by comparison gives reliability; throughput; zone based security and can be extended internationally for storage traffic
Unsure about the "extended internationally" bit for Infiniband (unless you're just meaning through use of multiple fibre links), but each of the other points is true for it.
It like Cobol and Fortran discussing who's best! Yes both are very used and useful still today... in but NICHE situations, in declining markets... not only in revenue, but declining faster in RELEVANCE.
No one cares. There is no significant money, innovation, or market impact from any FC or Infiniband improvements, because they have become .... irrelevant.
Storage is growing on NAS, iSCSI, well and if you ask Cisco... on FCoE too (joke alert) but the point is >> everything relevant is happening in another universe: the Ethernet universe. We don't need nor want more network protocols for us to manage, we need less!
Brocade and Mellanox might as well merge, so they can comfort one another ... together in their deathbed.
Yes, growing so much and for so many ages, that the stock topped at $114 four years ago, and now has been flat for 3 years at ~$50.
Ooooohh, but I guess you are the visionary, everyone else just fails to see it because...
And yes, sure Infiniband is about to change the world anytime soon too...
I just can't wait to hear it...
Explaining the ups and downs of the stock market is not always easy, but if you look at the Mellanox public earnings reports, you'll see that its been a story of continuous revenue growth since early 2013. Whether it's InfiniBand, Ethernet, or both, Mellanox has been growing nicely.
Disclosure: I am a Mellanox employee.
This post has been deleted by its author
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
