Ethernet vs. Fibre is the War Over? Cisco Announces New 9718 Director Class Switch

Ethernet will never completely replace Fibre Channel, they said – and they were right. Fibre Channel will always be faster, they said – and they were wrong. In fact, what they were wrong about was whether or not a war even existed. Maybe there was a time when Brocade ruled Fibre Channel (and didn’t do Ethernet) and Cisco ruled Ethernet (and didn’t do Fibre Channel). But once Cisco got into the Fibre Channel business and started selling arms to both sides of the war, many predicted it was the beginning of the end for Brocade. While that point is debatable, there is no doubt the pressure is on.

Gone are the days when one must decide between Ethernet and Fibre Channel (FC). Multi-protocol switches and ports take care of that. It is unclear why anyone would buy a switch that only supports Ethernet or FC because the future is simply unclear. Buying a switch – even an edge switch – that only supports a single protocol would seem like painting yourself into a corner. Make today’s protocol decision based on what you typically use in your data center, but connect it to a system that can change its mind as fast as your CIO changes theirs. “Future proofing” is the smart move.

A key to future proofing is to make sure the switching product has enough potential density to scale your data center into the future, also known as port density. Another is to make sure that there is enough scalable bandwidth to manage the potential load that a highly dense virtual infrastructure with an all-flash storage back end requires both now and into the future. While 16Gbs FC and especially 32Gbs FC seems like more than enough today, it may not be in the future. Particularly vulnerable are inter-switch links (ISLs), the ports that inter-connect directors to other directors or to other switches.

Cisco’s recent announcements are designed to meet these requirements in two ways. The first is the MDS 9718, a director with the highest port density in the industry – 768 ports in a single chassis. That many ports in a single chassis means you can connect far more hosts or edge switches into your core without wasting money on ISL links. Cisco claims to cite plenty of use cases that involve consolidating several of their competitors’ directors into a single MDS 9718 chassis. Fewer chassis means less wasted power, fewer things to manage, and fewer ISL links. They also announced that this is the first switch to allow you to program it using a native REST API – big news to DevOps and cloud environments.

Another unique capability is a 40 GbE FCoE interconnect link. Instead of using 16Gb FC as your interconnect, network designers can use one of these. Even if you have no FCoE in your environment, you can use Cisco’s multi-protocol switches to connect to each other using 40 GbE FCoE port, which is 194% faster than a 16Gb FC ISL. It’s even 47% faster than a 32Gb FC port.

StorageSwiss Take

Multiprotocol switches are a compelling alternative to their single protocol brethren and Cisco seems to be in a position to lead this product category. The unique use of a 40Gb FCoE ISL that is almost 200% faster than its closest competitor is a great example of Cisco continuing to innovate.

W. Curtis Preston (aka Mr. Backup) is an expert in backup & recovery systems; a space he has been working in since 1993. He has written three books on the subject, Backup & Recovery, Using SANs and NAS, and Unix Backup & Recovery. Mr. Preston is a writer and has spoken at hundreds of seminars and conferences around the world. Preston’s mission is to arm today’s IT managers with truly unbiased information about today’s storage industry and its products.

Tagged with: , , , , ,
Posted in Briefing Note
3 comments on “Ethernet vs. Fibre is the War Over? Cisco Announces New 9718 Director Class Switch
  1. Storage Bear says:

    Correction: 32Gb FC ports are available now, so it’s not a question of when. I notice however that you neglected to mention 128Gb FC, which is also part of the Gen 6 Fibre Channel standard, and also available now, at least at top rack. (Kind of takes the luster off of 40GbE, huh?) Frankly I’m surprised that Cisco would rally around 40GbE, a technology that’s clearly on it’s way out, at least at the server. Now had they chosen to go all in with the new 25GbE / 50GbE standard, *that* would have been interesting.

  2. wcurtispreston says:

    I’ll take that extra phrase off the 32Gb/s reference.

    As to 128 Gb FC, I only find that as a standard. I don’t see it in any actual products that I’ve looked at. Can you provide a reference to any such products?

    As to the “on it’s way out” comment, I’d say they all are. The point is that neither FC or Ethernet is dead, regardless of multiple attempts to say so.

  3. fstevenchalmers says:

    The answer, of course, to the question of Ethernet vs Fibre Channel is best considered not product by product or year by year, but as a progression over the course of decades.

    It takes a decade to establish a new technology by which servers talk to storage (and the vast majority of technologies brought to market never reach critical mass), and once one is established, even once something more fit for the role is available it takes decades for the established technology to fade. Fibre Channel earned its dominance in shared storage in the Enterprise, but did not win any design-ins in the cloud, nor in the go-back-to-using-the-storage-in-each-server pendulum swing Microsoft embarked on a decade ago. As much as my tangential role in the initial launch of Fibre Channel leaves me hoping for its immortality, reality is that Fibre Channel is now in the “decades to fade” part of the technology life cycle.

    The next era in storage will be characterized by reducing the server path length, and the latency, for access. It will break a lot more 50 year old assumptions than we did bringing Fibre Channel to market, and will stumble in worse ways. NVMe over Fabric is only the beginning…and is probably the last which will use a host stack as we know it. Next comes server resident, byte writable (not flash) persistent memory. It will take *real* software defined storage to turn that resource into real storage systems.

    In that context, Ethernet vs Fibre Channel to carry storage traffic which goes through the venerable SCSI stack on the host reminds me a lot of San Francisco’s old “blue and gold fleet” vs “red and white fleet” tour boats…which loudly competed but were actually owned by the same family.

Comments are closed.

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 22,215 other followers

Blog Stats
%d bloggers like this: