In its bi-annual SAN 5-year Forecast Report, Dell'Oro Group adjusted predictions for FC sales downward from its January forecast. The new forecast calls for Fibre Channel revenue from switches and host bus adapters (HBAs) of $2.36 billion this year, $2.68 billion in 2010 and $2.86 billion in 2011. Previously, Dell'Oro Group saw revenue of $2.47 billion this year and $2.7 billion next year.
The FCoE forecast is for revenue of $77 million in 2009, $274 million in 2010 and $583 million in 2011. According to Dell'Oro Group projections, Fibre Channel over Ethernet revenue will grow approximately $300 million between 2010 and 2011 vs. a $180 million increase for Fibre Channel.
Factors affecting FCoE adoption
Several factors will push FCoE by 2011, Dell'Oro said. Single-chip multiprotocol adapters hitting the market now could spur adoption, and Intel Corp.'s new Xeon Processor 5500 series (previously code-named Nehalem-EP) has the potential to deliver 50 GBps throughput in a commodity server, giving a big boost to 10 Gigabit Ethernet (10 GbE) -- a foundation of FCoE.
"Xeon 5500 servers change the playing field," Dell'Oro said, pointing out that 10 GbE adoption is generally considered the key prerequisite to FCoE adoption.
FCoE adapters and switches have been trickling into the market in recent years mostly from Brocade Communications Systems Inc., Cisco Systems Inc., Emulex Corp. and QLogic Corp. Last week, NetApp became the first storage vendor to offer end-to-end native FCoE connectivity on storage arrays.
Others expect a slower move to FCoE
Rob Stevenson, managing director of storage research at market research firm TheInfoPro Inc., said 10 GbE Ethernet is already becoming a reality in the enterprise data center at the core of the network. However, he said it's unclear what impact 10 GbE network-attached storage (NAS) and iSCSI will make on FCoE growth.
"Everyone I talk to believes FCoE will become the dominant core transport protocol," he said. "But I don't know if users will use it for everything, or if in three years there might be more targeted decisions at the application level."
Stevenson said storage shops are slowly moving to 8 Gbps FC, with FCoE still on the back burner. "The full refresh cycle to 8 Gbps Fibre Channel is happening at about half the pace of the transition from 2 Gbps to 4 Gbps," he said. "It'll take about two-and-a-half to three years to reach full maturation – about the same time as 10 Gigabit Ethernet."
At that point, organizations will face a choice between the next Fibre Channel refresh to 16 Gbps FC, or a mixture of Data Center Ethernet (DCE) and FCoE. Those who go with DCE will have a choice of NAS or iSCSI with 10 GbE and eventually 100 GbE. Stevenson said unstructured data is expected to outpace the growth of structured data during the same timeframe, which can push many people toward NAS.
Administrators seek alternatives to Fibre Channel
IT administrators are currently grappling with upgrade paths. Jeff Korte, senior systems administrator at a global telecommunications software and service provider, said his company is in the early stages of investigating FCoE. He expects the protocol to become "a legitimate enterprise alternative to Fibre Channel."
Korte, who asked that his company not be identified because of internal policy prohibiting him from representing it in the media, said he has approximately 160 TB in his data center and 3 PB of data globally.
"There are certain markets that require using proven technology and certain enterprises that have deep investments in traditional Fibre Channel infrastructure," he wrote in an email to SearchStorage.com. "However, as the price of 10Gbit Ethernet comes down, people are going to be looking for alternatives in 'less critical' environments that are more tolerant of configuration change."