Intel Pushes Fibre Channel Over Ethernet via Linux

Ethernet is on a march to become the dominant interconnect for all types of network traffic.

Although Ethernet is the dominant interconnect for LANs (define), when it comes to storage and SANs (define), it lags other technologies, notably Fibre Channel (define). That’s about to change.

Thanks to a new open source effort led by Intel, Fibre Channel over Ethernet (FCoE) technology is coming to Linux. FCoE may end up eliminating the gap between Fibre Channel and Ethernet providing users with more choice and flexibility in their networks.

“We think storage over Ethernet is a great idea,” Jordan Plawner, Intel’s storage planner and technologist, told “We’ve open-sourced the FCoE initiator, which is the first step to getting native OS support over time. We believe this will help accelerate adoption of FCoE.”

In storage terms an initiator is a key technology. Plawner explained that an initiator is code that establishes or initiates a session with a source target and helps to send data to the target. With iSCSI (define), which allows SCSCI traffic to go over Ethernet all the major OS vendors already have a native initiator stack so any server that ships is iSCSI ready.

“In the Fibre Channel world today there are various proprietary stacks from the hardware vendors and their can be issues around interoperability,”
Plawner said.

“In the Ethernet world we are used to having as much support natively in the operating system for whatever feature we’re talking about.
Our goal in the FCoE world is to make source code available to OS vendors and we’re stating with Linux so there can be a native FCoE stack.”

FCoE itself is also not a departure from core Fibre Channel in terms of its core data structure.

“The most important point about FCoE is that it doesn’t change the Fibre Channel frame,” Plawner explained. “You’re simply moving the Fibre Channel traffic onto a different fabric. You’re not taking apart the Fibre Channel frames.”

Since FCoE looks like regular Fibre Channel traffic, an enterprise’s security and traffic management infrastructure will work across FCoE fabric as well as the FC fabric.

“It becomes seamless to adopt and is very evolutionary and not revolutionary,” Plawner noted.

The FCoE
standard was proposed in April 2007
. Full ratification is expected sometime in 2008.

The decision to go Open Source with the FCoE initiator was a calculated strategic decision by Intel.

“We’re developing the code for our own products and we didn’t have to open source this. We could have just said by our solution,” Plawner said. “We took the more strategic view that it would benefit everybody especially the adoption of FCoE if we open source the code. So we’re picking up the social cost of improving and accelerating FCoE adoption.”

Plawner also admitted that by going Open Source, which can be used by anyone, the FCoE initiator benefits Intel’s competitors, too. That’s not necessarily a bad thing.

“I hope our competitors start helping us with the enablement effort,” Robert Love, Intel’s leader of the Open FCoE project, told

The FCoE effort is not intended to necessarily displace native Fibre Channel, iSCSI or other storage fabrics.

“It’s a matter of choice,” Plawner explained. “It provides another way for people to connect existing servers to Fibre Channel SANs. In iSCSI you have choice and we are planning to bring that same choice to the Fibre Channel world.”

Article courtesy of

Latest Articles

Follow Us On Social Media

Explore More