What Cisco's power play really means

Rabu, 18 Maret 2009 ·

Ever since Cisco's $150 million investment in VMware a few years back, the two companies have reportedly enjoyed a close relationship. But until the Cisco Unified Computing System announcement, little emerged in the way of collaborative efforts, a few virtualized switches and whatnot aside.

Some are hailing the Unified Computing System, which bundles a blade server with networking and server virtualization, as the Next Big Thing. But I find it rather underwhelming.

In the storage area of my lab, there sits a rarity: a 3Com Netserver. Never heard of it? Neither has anyone else. Released in the late 1990s, my model boasts a pair of 200MHz Pentium Pro CPUs, 128MB of RAM, and an odd 3Com SuperStack interface block. It was 3Com's entry into the fixed-purpose server market, designed to run RADIUS and other back-end tools to support 3Com analog dial products and other errata. It reached end-of-lif about a year or so later because commodity servers could do the job as well or better, were cheaper, and came without the proprietary ball and chain.

I see elements of the 3Com Netserver in the Cisco/VMware announcement. Cisco is probably going to be marketing and selling fixed-purpose blades, most likely manufactured by Quanta or another third party, branding them as Cisco devices, and trying to sell them as high-end virtualization platforms. They may have picked a good baseline architecture in the Intel Nehalem foundation, but otherwise, it's still just a blade server with lots of RAM and a tarted-up chassis. It's also likely to be somewhat restrictive when compared to general-purpose blade offerings from HP, Dell, Sun, and so forth. So why bother? Why would Cisco want to get into this relatively highly competitive market where it has no history?

The answer is that the company wants to force the standards it's developed on the rest of the industry. It wouldn't be the first time. I have a few Cisco 7960 Power-over-Ethernet phones that won't work with anything but a proprietary Cisco PoE switch because Cisco tried to force its PoE standard before the 802.3af standard was ratified. This left early adopters in the lurch, with phones and switches that don't work with anything else since they conform to nonexistent standards.

There's little technical detail to be gleaned from the announcement, but the big items are the implementation of VN-Link switching (which will essentially allow VMs to be controlled at the network layer like a physical server would be), aggregation of 10G switching within the chassis, Fibre Channel over IP, and a few other tidbits -- but they're all Cisco's standards, not the industry standards. These technologies are necessary steps along the virtualization development road, but it's generally a good idea to make sure the rest of the computing industry is playing the same tune, especially when you're talking about six-, seven-, and eight-figure investments for the early adopters.

Then there's the centralized management side of the announcement. Forgive me for being a bit leery of that, since Cisco hasn't exactly been a paragon of innovation in terms of management tools, utilities, and frameworks. Why is a networking company offering virtualization management? It's along the same lines as a taco stand offering tax advice.

Interestingly, Cisco's Datacenter blog has tried to preempt any criticism of the announcement (inexplicably referencing the iPod), so I'm certain that it's well aware of the potential for this to fall flat. I'm certainly wary of Cisco's ability to succeed in this market.

Cisco and VMware have said that there will be more detail -- including pricing and availability -- coming in the next few weeks, so perhaps some of my skepticism will become obsolete as more information arrives. I'd love to get more detail on, say, smarter switching for VMs, 10G storage and network aggregation, and so forth. Meanwhile, I'm not going to bet the farm that Cisco's and VMware's vision will triumph.

0 komentar:

ShoutMix chat widget