The goal of the Open-FCoE project: to encourage and support the development of a native Fibre Channel over Ethernet (FCoE) code for Linux, so organizations can leverage their Ethernet networks to connect existing servers to Fibre Channel SANs — without having to be locked into a particular vendor’s technology.
As Love wrote to colleagues, “FCoE will allow systems with an Ethernet adapter and a Fibre Channel Forwarder to login to a Fibre Channel fabric … [that] was previously reserved exclusively for Fibre Channel HBAs. This technology reduces complexity in the data center by aiding network convergence. It is targeted for 10Gps Ethernet NICs but will work on any Ethernet NIC supporting pause frames. Our code base provides a Fibre Channel protocol processing module as well as an Ethernet based transport module. The Open-FC module acts as an LLD [low level driver] for SCSI and the Open-FCoE transport uses net_device to send and receive packets.”
Open-FCoE Added to Linux Kernel
In April, Open-FCoE moved a step closer to its goal when its FCoE initiator, along with the data center bridging (DCB) drivers, which provide the quality of service that organizations need to be able to run Fibre Channel over Ethernet on a unified wire (along with other traffic types), were included in the 2.6.29 Linux kernel. The Open-FCoE project also released a generic Fibre Channel library and a generic Fibre Channel over Ethernet library, which provide a standard interface for Fibre Channel or FCoE devices to plug into the operating system, which Cisco (NASDAQ: CSCO) is using in one of its drivers.
As a result of being included in the Linux kernel, Open-FCoE’s code was also included in Red Hat’s (NYSE: RHT) Enterprise Linux (RHEL) 5.3 distribution as a tech preview, “indicating the future acceptance of this technology as part of the Linux infrastructure, much like iSCSI, TCP/IP and other native OS protocols,” said Jordan Plawner, senior product planner at Intel.
Plawner added that “based on community involvement, Linux Maintainer feedback and involvement by the Linux Distributors (RH and Novell), we expect a certified FCoE initiator with DCB support to ship in RHEL6 and SLES11-SP1 [SUSE Linux Enterprise Server 11], as well as other operating systems in 2010.”
That’s good news for SAN and network administrators, said Plawner, because it will give them “the option to standardize on a protocol provided and supported by the OS vendor, instead of choosing among the handful of proprietary initiators from unproven recent entrants into the Ethernet space.
“We believe fundamentally that the best way to do storage networking is to standardize on protocols and network stacks that are native in the operating system — that do not use proprietary stacks from independent hardware vendors or adapter vendors,” Plawner added.
And with about half of all storage networking today being done over Ethernet, FCoE provides “an opportunity to bring the simplicity of Ethernet to Fibre Channel while not polluting Ethernet with FC’s compatibility issues,” he said.
The Benefits of Open-FCoE
Addressing IT administrator concerns regarding FCoE, Love and Plawner pointed out that FCoE doesn’t change the Fibre Channel frame. “You’re simply moving the Fibre Channel traffic onto a different fabric,” said Plawner. So an enterprise’s security and traffic management infrastructure should work the same across FCoE as it does across regular Fibre Channel. Moreover, FCoE can help enterprises consolidate their I/O, cables and adapters.
As for who stands to benefit the most from FCoE, right now FCoE is targeted at data centers that use 10 Gigabit Ethernet. “We wanted FCoE to be native in the Linux kernel, so you can just take our 10 Gigabit driver that you’re using for normal Ethernet TCP/IP and now you’ve got Fibre Channel over Ethernet as well, at no additional cost,” said Love, who is responsible for maintaining Open-FCoE.org.
Sun’s Open Source FCoE
As for who is actually using Open-FCoE’s stack, it’s hard to know exactly, said Love and Plawner, because the code is part of the Linux kernel so anyone who has access to the Linux kernel can use the code. And Intel is not the only organization working on an open FCoE solution. In August, Sun’s (NASDAQ: JAVA) OpenSolaris project released its own FCoE Initiator and FCoE Target. And several other storage vendors have also been jumping on the FCoE bandwagon too.
NetApp (NASDAQ: NTAP), for instance, recently made news by becoming the first storage vendor to offer native end-to-end Fibre Channel over Ethernet, in its FAS and V series storage systems, using QLogic (NASDAQ: QLGC) converged network adapters (CNAs) and Brocade (NASDAQ: BRCD) switches.
However, while analysts and vendors are predicting a bright future for FCoE, a lot of the adoption numbers currently being bandied about may be overly optimistic, said Gordon Haff, a senior analyst at Illuminata.
Plawner agreed. “I’ve seen market research reports telling me how many FCoE ports were shipped in 2008, and I absolutely know those numbers were inflated 300 to 400 percent,” he said.
Still, despite the hype, the Dell’Oro Group recently predicted that by 2011, FCoE will have made significant in roads at enterprises. And if Intel has any say in the matter, new servers will have open source FCoE inside.
Jennifer Lonoff Schiff writes about IT issues and is a regular contributor to Enterprise Storage Forum.
Follow Enterprise Storage Forum on Twitter