iSCSI: Game over
The FCoE killing fields
NetApp, Cisco, QLogic, Emulex and VMware have all proclaimed FCoE (fibre channel over Ethernet) heaven is just around the corner. So is this the end for the iSCSI internet storage standard? Has Dell's EqualLogic purchase suddenly been devalued?
The thinking goes like this: iSCSI sends SCSI commands wrapped in TCP/IP to a storage array across an Ethernet wire, typically a 1Gbit/s line. This provides block-level access to a shared storage resource, an IP storage area network (SAN).
It's much less expensive than a traditional Fibre Channel (FC) SAN as you don't need any FC switches sitting between the accessing servers and the storage arrays. That means no FC cables and no FC skills in the data centre. The trade-off is that Ethernet has a longer latency than FC and can lose packets causing a retransmission, meaning even more latency.
FC is also faster, currently running at 4Gbit/s and transitioning to 8Gbit/s. Eighteen months ago the idea of sending FC packets across an Ethernet link (FCoE) instead of a FC fabric surfaced. Ethernet would unify all networked storage access by being used for file-level network-accessed storage (NAS) and block-level SAN access using either iSCSI or FC protocols.
Terrific. FCoE meant you could enjoy FC speed - using 10gig Ethernet - and predictable latency with no packet loss. You could evolve an existing FC SAN to include FCoE links and, over time, discard your FC fabric, replacing it with an Ethernet infrastructure - much, much cheaper with large cable cost-savings.
So what do you need iSCSI for, then?
But... why do you then continue to need iSCSI? It evolved to solve the FC fabric complexity and cost problem. FCoE makes that go away by turning Ethernet into a virtual Fibre Channel.
FCoE product has sprung up: Cisco has its Nexus 5020 switch which EMC is reselling. So too is NetApp, which has added a native FCoE interface to its FAS (fabric attached storage) and V series arrays, courtesy of a QLogic FCoE target card.
There's more - QLogic and Emulex have FCOE converged network adapters (CNAs) certified by Cisco. Brocade is supportive of FCoE. EMC's Clariion CX4 can have a FCoE interface added in short order. End-to-end FCoE storage area networks have just become possible.
Another thing - Ethernet is transitioning to 10Gbit/s and a new Data Centre Ethernet (DCE) standard will provide predictable latency and not lose packets. FCoE should be a T11-ratified standard by around the middle of next year.
It's true that a 1Gbit iSCSI link is cheaper than a 10Gbit/s FCoE link. The iSCSI link needs a network interface card (NIC) - hopefully with a TCP/IP offload engine (TOE) on it. This saves the server processor from having to spend cycles formatting all the TCP/IP stuff, which is complex.
FCoE requires a CNA that adds the fibre channel to Ethernet. Currently an iSCSI 1gigE line plus TOE card is less expensive than a 1gigE line plus CNA. Let's wait 18 months to two years and see both 10gigE and CNA costs come down as well as the need for SAN access speed to go up, because you're running more virtual machines in your faster multi-core servers. Then wouldn't it be easier to junk iSCSI and jump wholescale to FCoE?
You would save cost and have one less network protocol to think about. QLogic's Roger Klein, a VP for its host solutions group, thinks this is a very likely scenario.
Xiotech's chief technology officer, Steve Sicola, not only thinks this is the way things will play out but that it could substantially reduce the value of Dell's EqualLogic acquisition. Dell spent $1.4bn buying EqualLogic's iSCSI market technology late last year and now the whole iSCSI market looks set to be killed by FCoE. Bad move, guys.
FCoE levels the block-level access playing field. There's no longer the marketing distinction between a Fibre Channel array - fast, costs more - and an iSCSI array - slower, less reliable access, cheaper. It's all Fibre Channel. The distinction will move to the media inside the array and the array software. So all iSCSI array vendors will need to rethink their marketing propositions and retool their product strategies.
There is no way out for iSCSI. It's at war and FCoE is the killing field. ®
Sponsored: Hyper-scale data management