I need to clarify that while I understand why the Ethernet Technology Summit did not succeed, I still am extremely disappointed in its demise.  I believe the Ethernet Technology Summit is an excellent model for a show.  The reality is that we don’t need so many different technologies to implement a storage or communications network.  It is possible – and efficient and cost-effective – to use Ethernet end-to-end.  In fact, Ethernet is clearly the best candidate for enabling the converged network across applications. 

Many industry shows and conferences focus on applications.  The strength of the Ethernet Technology Summit is that it was going to focus on Ethernet.  It would have also brought all the various flavors and forms of Ethernet together in one show.  This would be good for the industry, for many reasons.  It is a shame that the timing of the economic downturn made it economically unsound for its backers to proceed with the Ethernet Technology Summit.  This conference would have: 

• Provided a forum for developing a single language across Ethernet applications
• Brought engineers and network administrators together to share needs and requirements
• Enabled all interested parties to work together to make Ethernet technology as a whole more robust for all applications using Ethernet
• Highlighted the role and capabilities of Ethernet to network administrators, demonstrating to them the feasibility and reliability of how an end-to-end Ethernet implementation could meet their specific needs. 

When times get better, we should revisit this idea and give it another try.

The Ethernet Technology Summit was slated to be the first industry conference dedicated to just Ethernet throughout the networking space.  As a conference, it was intended to focus on new emerging Ethernet technologies in LAN, data centers, telecommunications, and high speed applications, to name a few, as well as serve as the industry’s first gathering for dedicated Ethernet discussion from data to communications to storage, from short to long distance, from node to core, and on and on.  It was a great idea to bring together discussions on Ethernet given that Ethernet technology is the best candidate for network convergence end-to-end. At other shows, like OFC or SuperComm, Ethernet is portrayed as a strong supporting co-star working with other technologies.  These shows are more application-focused, and so the ubiquity of Ethernet is seen through its presence at each of these events.  However, Ethernet’s dominance in the industry, as well as its ability to serve as a reliable end-to-end technology in many applications, can be easily overlooked. 

This is perhaps one of the key reasons for the Ethernet Technology Summit’s failure to launch.  With Ethernet already a major consideration at every other industry conference, many people might have felt that they didn’t need to also attend an Ethernet-only show.  In addition, the downturn took its toll on projected attendance, with companies cutting even core shows from their must-attend list.  If companies are deciding to forego OFC, for example, what chance does a startup show have of catching their attention?  It might have also been the Ethernet Technology Summit’s focus on technology that failed to attract more attendees.  After all, most people are used to the application-based show model. 

It’s a shame the Ethernet Technology Summit has been cancelled.  But this is more a statement of the economic climate and biases of our industry than about Ethernet.  Ethernet is as strong as ever, with engineers working around the world to further increase its reach, including its continued progress towards widespread deployment of FCoE. 

At Supercomputing 2008 (November 18-20, Austin, Texas), nearly a dozen communications leaders came together to show deployment of several types of mixed traffic over a multi-vendor 10 GbE network infrastructure.  Dubbed the 10 Gigabit Ethernet Converged Data Center Demonstration, this event hosted by the Ethernet Alliance showcased how 10 GbE optical and copper technologies could interoperate and interconnect a complete data center.  Equipment included PHYs, NICs, CNAs, layer 2 and 3 switches, cabling systems, storage, test systems, and management consoles.  Test data ranged from application-to-application traffic, low latency streaming video from iSCSI storage, converged traffic types via virtualized servers, and FCoE over a lossless Ethernet network.

The importance of this event is that it shows that the Ethernet networking industry is working diligently to extend their hand to storage, so to speak, with as much vigor as the Fibre Channel industry is reaching out to the LAN.  The Data Center Ethernet (DCE) demonstrated enables the data center to carry many types of traffic, and FCoE will be the bridge that brings native FC data to DCE, leading to the Converged Data Center.  Put another way, DCE opens the door that is required to enable FCoE to serve as the media that will bring storage to the data center.

With DCE available, the Converged Data Center is that much closer to reality.  Several switch vendors in the industry claim that they can carry all types of traffic, and they have proven this in terms of transport.  They even have Priority Flow-based Control (PFC) mechanisms in place, making their Ethernet lossless so it can carry Fibre Channel traffic reliably while meeting the high standards associated with Fibre Channel for data integrity.

However, to be unbiased, we must admit the reality that not all of the pieces necessary for FCoE are in place yet.  This demonstration has shown that individual network components are capable of managing PFC to achieve lossless Ethernet.  But networks are not islands of individual nodes.  Storage networks are managed as an entity themselves, and reliable, lossless connectivity must extend across the entire network.  Specifically, congestion management has yet to be addressed.  The network must be able to interoperate and recover from failures as a whole as well as the individual components do.  If not, the data center cannot offer the same reliability as the storage network.This demonstration is wonderful news for FCoE.  Lossless Ethernet is still at a preliminary stage, and while there is yet much work to be done to enable FCoE in real-world networks, the momentum forward continues to increase.  With players like Cisco, Extreme Networks, Force10 Networks, Fulcrum Microsystems, Ixia, Mellanox, Panduit, Solarflare, Teranetics, Tyco Electronics, and QLogic paving the way, it is only a matter of time before many others follow to join in bringing FCoE to reality.

For specific information on the Supercomputing 2008 – 10 Gigabit Ethernet Converged Data Center Demonstration, including detailed descriptions of the data center configuration and test traffic, visit: http://www.ethernetalliance.org/images/SC08%20White%20paper.pdf.

For more information on the Ethernet Alliance, visit: http://www.ethernetalliance.org.

Last month the industry’s leading Fibre Channel vendors came together at the University of New Hampshire Interoperability Lab (UNH-IOL) to complete interoperability testing for up and coming FCoE products and solutions.  Device and storage system manufacturers participating in the FCoE Plugfest included Amphenol, Brocade, Cisco Systems, Emulex, Hewlett-Packard, IBM, Intel Corp, LSI, Mellanox Technologies, Molex, NetApp, and QLogic Corporation.  The Plugfest was sponsored by the Fibre Channel Industry Association (FCIA).

Finisar was a proud participant at the Plugfest, bringing a team of 9 engineers including development engineers, protocol experts, and networking experts to assist in interoperability testing using three of Finisar’s Xgig 10GE, 8G Fibre Channel Analyzers and Medusa Labs Test Tool Suite.  (More details outlining Xgig’s complete lineup of protocol-aware FCoE monitoring and analysis capabilities, including full-rate Traffic Generation, will be announced later this month.)

Finisar was literally in the middle of the Plugfest, sitting between test units as it monitored links and provided real-time results for every single interoperability test performed.  From our unique position, we watched as every FCoE initiator was found capable of communicating across the FCoE switches to both FCoE and Fibre Channel targets.  As participants ran into issues, they were able to fully characterize root causes and conduct on-site troubleshooting with the help of our protocol experts. With this information, they could apply quick fixes and proceed to the next level of testing.

The outstanding success of the Plugfest, with vendors achieving interoperability across a great many configurations, is exciting.  It represents another milestone towards the development and verification of the stability and interoperability of multi-vendor FCoE-based networks.  Clearly, the momentum behind FCoE is significant, as the results of the Plugfest show.  Such a high level of commitment from key Fibre Channel vendors will lead to rapid product availability and subsequent deployment as the standard continues its transformation from a paper spec to physical reality.
Some results from the Plugfest:

FCIA: http://www.fibrechannel.org/NEWS/fcia080929.html

Finisar: http://investor.finisar.com/releasedetail.cfm?ReleaseID=334685

Emulex: http://www.emulex.com/press/2008/0923-01.jsp

QLogic Corporation: http://biz.yahoo.com/pz/080922/150776.html

Brocade’s announcement of its intention to acquire Foundry Networks for $3 billion has sparked much discussion in terms of its impact on the FCoE market.  Michael Morris, writing for Network World, admits to questioning Brocade’s motives for wanting to enter the highly commoditized LAN switching market when Cisco already dominates it.  His conclusion is that Brocade may be attempting to look for differentiation through FCoE as the SAN is comprised of expensive FC ports whose profitability could be eviscerated if FCoE succeeds as expected.  Enterprise Strategy Group analyst Bob Laliberte believes the acquisition “instantly gives them the credibility, culturally, in large enterprises.”

The official stand at Brocade is that the acquisition of Foundry Networks was not motivated by FCoE but rather a move to extend their line of products and build on the synergy between their different technologies and markets.  Brocade has had a roadmap for FCoE in place, so it’s still not clear how the company will integrate Foundry’s technologies.  What is clear is that the FCoE market opportunities for Brocade will be large and of high strategic importance. 

Disruptive Technology

While Brocade will expand its product offerings and market reach, the acquisition of Foundry may not result in much of an immediate market shift given that Foundry has been considered a niche player.  However, the acquisition of Foundry will boost Brocade’s FCoE positioning at these critical, early stages of the FCoE deployment within the data center. 

Consider that convergence within the data center continues to lead to divergence in product requirements as standards evolve and become more complex, such as Enhanced Ethernet for FCoE and continued pushing for Power over Ethernet (PoE), QoS, and more robust security for VoIP and WLAN applications.  The opportunities for specialization lead to more niche markets and more openings for smaller players to gain localized dominance.

While FCoE as an emerging market could be considered a niche, early dominance of market share could expand to a long-term leadership position in the evolving SAN market.  This is possible because FCoE is clearly being considered as a disruptive technology, one that could change the underlying SAN infrastructure, business models, and, as a consequence, major players servicing and driving the industry.

This disruption may have a greater impact than current competition for existing markets.  Brocade has already gained a dominate position in the SAN switch market.  But from a technological perspective, some experts believe that developing for FCoE will not be as simple as sending Fibre Channel packets over an Ethernet link, but be like developing for an altogether new technology. The converged network of the future will bring many new use-cases to the switch fabric that was once limited only to storage traffic. And while the technology that Foundry will bring to Brocade isn’t the Enhanced Ethernet FCoE calls for, having expertise in Ethernet will be a critical requirement in the future success of FCoE solutions.

Put another way, no one yet has dominance in the FCoE space. 

What Brocade will do with the technology they acquire through the acquisition of Foundry is still a matter of speculation.  What is not in question is the growing momentum of FCoE and the continued support and investment of key industry players.

References:

1. Brocade Buying Foundry…Has to be an FCoE Play
http://www.networkworld.com/community/node/30200

2. Brocade + Foundry = Trouble for Cisco?
http://www.networkworld.com/community/node/30187

3. Brocade to buy Foundry for $3 billion
http://www.networkworld.com/news/2008/072108-brocade-to-buy-foundry-for.html?nwwpkg=brofound?ap1=rcb

4. Brocade-Foundry deal can’t hide majors’ dominance
http://www.networkworld.com/news/2008/072208-brocade-foundry-deal-cant-hide-majors.html?inform?ap1=rcb

5. Brocade’s Foundry buy will boost Fibre Channel over Ethernet market
http://www.networkworld.com/news/2008/072208-brocade-foundry-fcoe.html?page=1

6. Dell Oro Group take on Foundry and Brocade merger: Datacenter requirements are evolving towards converged fabric
http://www.networkworld.com/community/node/30262

Not all new standards are created equal. While each must pass through various design, ratification, and adoption phases, some standards penetrate the market more quickly than others. Rate of success depends upon many factors, including how compelling the technology is, how quickly reliable hardware and its supporting software are widely available, and how decision makers view the emerging technology.

FCoE is clearly compelling. It promises to:

• Converge Ethernet networking and Fibre Channel storage through a single, unified fabric
• Increase performance while simplifying management complexity
• Extend the reach and capabilities of the SAN while protecting existing infrastructure investment
• Provide more computing power in a smaller footprint, meaning fewer servers, less cabling, and more efficient power consumption.

Reliable hardware and software are already available and, in some cases, already installed:

• Cisco’s 10 GbE Nexus 5000 series switches will allow users to enable FCoE through software once the relevant standards have been ratified
• Emulex, QLogic and others have converged network adapters (CNAs) that support FCoE
• Intel has announced FCoE support for its 10 GbE server adapters

Alas, as is commonly the case with emerging technologies, the most critical element affecting FCoE adoption will be how the market accepts it. The human perception of FCoE’s readiness, more so than its actual technical viability, is what will determine when FCoE goes live. Several factors hindering FCoE include:

• Fear. Uncertainty about the economy is a key driver for prompting network administrators to “play it safe”. But playing it safe is too frequently a euphemism for spending too on soon-to-be-legacy technologies. If the economy is facing a downturn, FCoE can bring significant cost savings when they are needed most. The companies that recognize this will fare any hard times the best.
• Fear. FCoE brings together networking and storage teams, organizations that typically have had to compete with each other for resources. Convergence may be viewed as impending obsolescence and a threat to job security. The very people who stand to gain the most from embracing FCoE may feel rather that they have the most to lose.
• Fear. FCoE is an emerging standard and, as with any new technology, it represents gain as well as risk. This type of fear is rooted in lack of familiarity. Risk always seems higher with those things we don’t yet understand. Expect long proof-of-concept trails before the skeptics are convinced this is ready for their data center.

Excellent process is being made in the design and deployment of reliable FCoE equipment. However, for FCoE to become one of the standards that reaches market faster than others, those companies invested in its success must take a firm hand in overcoming the fear that threatens to forestall its advantages. Clearly, counting on the industry to accept FCoE because it’s a good idea is not enough.

• What other fear factors threaten to delay FCoE?
• What can the industry as a whole do to address these fears?

Further Reading: Infostor – “The Promise of FCoE, the Reality of Adoption”