Cisco Expands Its Nexus to Partner Blades

Cisco is now bringing its effort to create a unified fabric for regular LAN and storage SAN traffic to blade servers. The basic idea is to reduce the number of switches that a blade server requires in order to access both data and storage networks.

The new Nexus 4000 switch is an expansion of Cisco’s (NASDAQ:CSCO) Nexus unified fabric switch which first debuted with the Nexus 7000 switch in January of 2008. While the Nexus 7000 and its smaller cousin the Nexus 5000 are designed for regular servers and routers, the Nexus 4000 is specifically being designed for the needs of Cisco’s OEM blade server customers.

“What we’re looking to do is to build blade switches that are all 10 Gigabit and bring it to our partners’ blade servers chassis,” Ranjeet Sudan, Cisco senior product manager at Cisco told “The first stage of unified fabric we see is really about integrating SAN and LAN traffic across the server access link, whether it’s a rack mount server or blade server. We’ve done it for our partners rack mount servers, we’ve done it for Cisco’s own UCS server and now we’re talking about doing it for our partners.”

Cisco’s Unified Computing System (UCS) uses a Nexus 5000 switch to provide a unified fabric network link for the underlying servers.

Sudan said that with the increasing use of server virtualization, overall data center traffic is going up which is driving a need for larger SANs. The increased utilization of blade servers together with virtualization is also driving up the number of network interfaces that blade servers often require.

According to Sudan, blade server chassis can have 14 to 16 blades, where the servers are at the front of the chassis and the blade switches go into I/O slots at the rear of the chassis. The idea with the Nexus 4000 is to help lower the total cost of ownership through consolidation of those switches that fit in at the back of a blade server chassis.

“When we take a look at consolidation, we’re not only talking about consolidating the number of LAN and SAN switches, but actually greatly reducing the number of switches required in the chassis,” Sudan said.

Fewer switches to manage

He added that today, many users that are virtualizing their blade servers are deploying four to six network Ethernet interfaces per server with an additional two fiber channel interfaces. In total, there could be as many as six to eight switches that a typical blade server user might need to deploy. According to Sudan, with Nexus 4000 the total number of switches will be cut down to only two switches.

“That’s a huge reduction in the number of switches that you need to manage,” Sudan said. “It’s a reduction in the power budget you need to support as well as a reduction in the number of cables.”

Cisco has been manufacturing Catalyst blade server switches for IBM, HP and Dell for several years. Sudan said that specific announcements from partners would be coming in the future about the full technical specifications for each partner blade system.

Competition for Cisco is already in the market from some of those same partners. HP recently debuted its own blade switch, the ProCurve 6120XG, in an effort to provide an alternative choice for HP c-class BladeSystem users.

Cisco Catalyst blade server switches for IBM, HP and Dell servers are not currently being offered a trade-in program to upgrade to the new Nexus 4000. Sudan said that though Cisco doesn’t have a trade-in program today, it’s not something they wouldn’t consider in the future.

Overall, Cisco’s selling proposition is that a move to Nexus pays off.

“Generally the economics of moving to unified fabric, especially at the access layer are so compelling that we find that most customers that look at the trade offs, and even without a trade-in program, it makes sense for them to make the move,” Bob Nussbaum product manager at Cisco said.

News Around the Web