Among storage vendors, NetApp has been the most aggressive in its support of Fibre Channel over Ethernet (FCoE), claiming to be the first to offer "native FCoE," which it defines as end-to-end FCoE from the storage to the switch to the host server environment.
NetApp's high-end FAS 6000 series, midrange FAS 3100 series, FAS 3040 and 3070, and low-end FAS 2050 all support FCoE through a QLogic unified target adapter. The add-on adapter cards, each of which has two ports, plug into open PCI Express slots in the controller.
Which vendors are supporting FCoE?
Generation 1 unified target adapters, which became available in February 2009, were generally restricted to pilots and proofs of concept. The Generation 2 cards, which started shipping in September 2009, can support not only FCoE but also iSCSI through a product variance request. Next year, plans call for the addition of network-attached storage (NAS) support, according to Michael McNamara, a senior manager of product marketing at NetApp.
"With 10 Gig, there's a price premium, and it's going to take a little while for the Fibre Channel over Ethernet market to evolve and for end-user adoption to really take off," McNamara said.
NetApp claims to have customers who are already running end-to-end FCoE through Cisco's Nexus 5000 Series switches. But the architectural approach -- going FCoE from the converged network adapter (CNAs) to the top-of-rack switch to the storage -- offers limited scalability in the absence of core switches that support FCoE.
McNamara acknowledged that many larger customers prefer using core switches for additional flexibility and reliability, and he noted that NetApp expects some customers to delay production use of FCoE until core switches, such as the Nexus 7000, support FCoE.
Other major storage vendors expect to start rolling out FCoE support over the coming months. EMC Corp., for instance, plans to offer native FCoE ports on its Clariion array in 2010. Customers who buy a Clariion CX4 today will be able to upgrade to FCoE, according to Stuart Miniman, a technologist in the office of the CTO at EMC.
Miniman said EMC has been in no rush to support FCoE in its storage, since the top-of-rack switch-to-storage approach holds little appeal to customers. "Most customers want to have some aggregation typically through a director before they get to the storage, and I can't do that natively with FCoE today," he said.
Director, or core switches, don't yet support FCoE. Plus, the standards that will enable FCoE between multiple FCoE Forwarder (FCF) switches are in development, so users also currently are not able to go from one top-of-rack switch to another top-of-rack switch with FCoE traffic, Miniman noted. "I can't cascade them," he said. "I can only go to one switch and then from that switch, I have to switch over to Fibre Channel."
Users can, however, employ intermediate Ethernet switches that support the FCoE Initialization Protocol (FIP) to go switch-to-switch with FCoE traffic. Cisco's recently announced FIP-capable Nexus 4000 blade switch is an example, in connection with the vendor's Nexus 5000 Series top-of-rack switch, or FCF.
In other words, users won't be able to go end-to-end FCoE from a Brocade Communications Systems Inc. 8000 or Cisco Systems Inc. Nexus 5000 Series top-of-rack switch to a core switch or blade (such as Brocade's DCX or Cisco's core offerings) to storage, until the standards shake out. The "routing bridges" standard that will enable the FCoE switch-to-switch scenario is expected to be approved by the third quarter of 2010, according to Brocade's Zamer.
"Due to the state-of-the-art of FCoE technology, the proposed 'two hops' path is not possible on any FCoE product available in the market today," Zamer wrote in an email to SearchStorage.com. "All current products are first hop (single hop) products, therefore one cannot create a two hop (Brocade 8000 to blade, or Nexus 5000 to blade), because that constitutes a two-hop solution, which requires routing bridges (RBridge) that are capable of multi-hops."
This was first published in November 2009