The Great Cisco Switch
Join the DZone community and get the full member experience.Join For Free
The other day it unveiled what it thinks is its most important product in, oh, goodness, 15 years.
It’s a switch…a data center switch…a very big switch with incredibly massive bandwidth capacity that Cisco created for what it sees as the next-generation virtualized data center, its Data Center 3.0 vision.
It says the device, the Nexus 7000, 20 times the bandwidth of anything currently available, can copy the entire Wikipedia database in 10 milliseconds – a tenth the blink of an eye; download every one of the 90,000 movies on Netflix in 38.4 seconds; copy the entire searchable Internet in 7.5 minutes; run five million concurrent videoconferences; and add a web server in nine seconds, not your usual 90-180 days.
In techno-speak Cisco says Nexus can deliver 15 terabits a second of switching capacity and support 512 10-Gbps Ethernet, with the promise of 40- and 100-Gbps Ethernet.
The current top-of-the-line Cisco switch is good for two terabits of data a second.
But the real trick is Cisco’s unified fabric architecture combining Ethernet, IP and storage capabilities. That way, with a smarter kind of Ethernet, Cisco is promising that all servers will have access to all network and storage resources – even if they’re on the other side of the globe.
It’s the enabler of the grander kind of supra-territorial consolidation and virtualization that Cisco has in mind. Distance simply won’t matter and no more of this unused computing capacity.
Cisco’s unified fabric eliminates parallel storage and computational networks, and reduces the number of server interfaces as well as the cabling and switching infrastructure. It means higher-density server form factors and increased data center workload output – and, combined with virtualization, can save power.
Cisco estimates an 8% cut in power consumption with the unified fabric, maybe $20 million over the life of a 20- or 30-megawatt data center.
And thanks to the new NX-OS operating system, Cisco is confident in promising zero service-disruption upgrades in production systems.
Cisco has spent three years and $250 million developing the thing, a sort of hypervisor for the network capable of creating multiple network instances across a single fabric.
According to the Wall Street Journal, it’s registered an amazing 1,500 Nexus patents.
And it hasn’t work it all out yet. Cisco says “key components of the unified fabric architecture include unified I/O interfaces and Fibre Channel over Ethernet support to be delivered in the future.”
Cisco figures it’s early and is anticipating a long sales cycle if for no other reason than it’s pricey. The switch starts at $75,000 and your average installation will probably run $200,000, making it a big boys’ switch, a natural, Cisco thinks, for the data video and collaborative applications crowds, the financial institutions and search firms.
Nexus, which comes in a three-and-a-half-foot box, is being tested by Microsoft and the Lawrence Livermore National Lab and isn’t supposed to be generally available until the second quarter. Cisco is, however, taking orders.
Meanwhile, Cisco said Dell will be using its new – but less visionary – Catalyst Blades Switches in its new PowerEdge M1000e blade server enclosure.
There is Virtual Blade Switch (VBS) technology on the Blade Switch that allows up to eight switches to be managed as one logical switch.
VBS is supposed to provide 160 Gbps upstream performance and double bandwidth to the server at the same time, something Web 2.0 applications should appreciate.
Opinions expressed by DZone contributors are their own.