Cisco Takes On Storage

Cisco has had a few years to work its magic in the storage market, so when we got the opportunity to speak with Jackie Ross, VP of marketing, and Rajeev Bhardwaj, senior product manager in the Cisco storage group, we asked them some hard questions about the issues paramount in the storage industry.

Not surprisingly, Cisco comes across as rather proud of its achievements to date--it's been growing storage-switch market share at a rate that should be making Brocade, McData and QLogic take notice. The company also has managed a reasonably steady stream of enhancements to its storage-switch gear that it clearly hopes will generate growth quarter over quarter.

But we wanted to know how Cisco plans to address the big questions that confront storage managers today--competition to expensive Fibre Channel and SCSI gear from a new breed of low-cost SAN and NAS products based on Windows Storage Server or iSCSI; the iSCSI versus FC debate; and the challenges of making disparate architectures play nicely together. Coming at the market tangentially are blade servers and increasing storage speeds. Our questions hit on all these issues, and also delved into Cisco's plans and how customer feedback impacts its direction.

Our overall takeaway from Bhardwaj and Ross: Cisco is addressing all these issues from the perspective of giving users choice.

To make certain we all understand the issues, here's some background:• Fibre Channel is a short-run storage protocol that runs over optical cable. It's used for connecting storage to servers in traditional SANs and for connecting storage to SAN controllers.

• FCIP tunnels Fibre Channel commands over an IP network, giving you FC functionality without the short-run and optical limitations.

• Finally, iSCSI is similar to FCIP but tunnels SCSI disk commands over IP. The iSCSI market has evolved such that the device normally on the other end is not just a hard disk, but a disk array.

All these technologies make remote storage appear as a local drive, but performance varies--any IP-based storage is slowed by network bandwidth and the requirements to package and unpackage data, while FC is limited by requiring specialized hardware at each end. Cisco's position: FC will continue to grow; FCIP will be used to interconnect SANs and simplify backups; and iSCSI will be useful at the low end. We question whether, long term, FC has legs and whether FCIP will gain popularity.

Still, Cisco's platform is hard to beat if your crystal ball tells you the only constant in the storage space--at least for the next several years--will be change. Although we believe IP-based storage will win out eventually, other industry pundits say iSCSI will never be high-performance enough to compete with Fibre Channel directly. Cisco has deftly managed to appease both sides with its 9000 Series line of SAN switches. If you need FC today but require FCIP access to a remote SAN or consider iSCSI a serious threat to FC, then Cisco has a box for you: the 9216i. Just call it Switzerland."We're protocol agnostic--whatever the customer requires," Ross says.

And when technology changes your requirements, you slip in a new blade or change a configuration setting and run cables. We've tested some of these new Cisco products that handle FCIP, iSCSI and FC, all in a single box. Our impression: Although you won't use FCIP in this environment unless you're already an FC shop, and it is a little pricey, the 9216i does a good job of tying FC with IP storage protocols.

We disagree with Cisco's assertion that the market for FCIP in the enterprise is larger than most analysts and editors give it credit for, but if you're running FC and need remote access--say, for interconnecting SANs, accessing a SAN from remote servers (such as remote departmental servers requiring access to your data-center SAN) or for backups--interoperability is useful. And because Cisco supports FCIP in addition to FC, iSCSI and even mainframe storage connectivity through FiCon (fiber connectivity), it doesn't hurt iSCSI or FC users to have FCIP built in, and it's probably a selling point at larger organizations.

Cisco also believes that the features and perceived quality it has applied to data networking is a huge plus for its positioning in the storage networking arena. We have to agree. Ross characterizes early adoption of SAN technology as on par with early adoption of TCP/IP, and puts forth the premise that Cisco's background gives it a leg up on developing advanced storage-networking features. If VSAN, the ability to make SAN-based VLANs to segment data traffic, and inter-VSAN routing capabilities are any indication, Ross is right. As more storage networking heads into the IP arena, this strength should serve Cisco well.

We told Ross and Bhardwaj that though Cisco's traditional market is oriented toward the enterprise, where FC is the ruling technology, the up-and-coming IP storage market is going to grow the most at the low end, where small and midsize businesses don't want to have to learn IOS and FC to set up storage. In response, they pointed us to Cisco's GUI, which Bhardwaj says does the same things as IOS.Bhardwaj also directed our attention to the fact that growth in the low end is unlikely to be FCIP--that will be used to hook remote servers to SANs and to handle SAN-to-SAN linking over a WAN or MAN. Rather, iSCSI will be the system of choice for low-end users, and with its GUI, iSCSI is relatively easy to configure.

One area in which we completely endorse Cisco's approach to the low end is in education of IT staff before implementing storage networking. As smaller shops begin to use more SAN technology, it would be nice if they could architect a long-term plan instead of throwing a NAS or iSCSI device on the network and assuming they're done. When these shops outgrow the capacity of that device, they'll need to do something more complex, and if the users have training beforehand, they can make decisions that will work for them now and in the future.

Come Together

When we asked about the convergence of IP and SAN traffic and support for blade servers, Bhardwaj and Ross again pointed out Cisco's successes in both the IP and SAN markets as proof that it's on top of convergence. Ross added that the company already supports the unique storage needs that blade servers meet through partnerships with blade-server vendors, including IBM.

Overall, we feel Cisco is on the ball in the storage market. It has the background and the talent to compete in an arena that is becoming increasingly network-oriented. It also has a ready-made market, with its dominant installed base of data networking gear. And because the company supports your SAN regardless of which protocol you choose to implement, it's also appealing for merger and acquisition purposes.However, despite its outward endorsement of FC, we see no evidence that Cisco has truly changed its belief that IP networking is the wave of the future for SANs. It has segmented the market such that it has a good idea of what is needed today and what will be needed in the future: Short term, it's selling FCIP for inter-SAN connectivity, and iSCSI to hook departmental machines up to the SAN. Longer term, it's looking at iSCSI for small and midsize businesses, and a mix of FC-iSCSI-FCIP for enterprises. This is a sound plan evidencing a good understanding of the technologies in question.