Cloud Computing: Cisco Boss Makes Contradictory Assumptions About Virtualization

Grazed from NetworkComputing. Author: Kevin Fogarty.

During the past decade, nearly everything in the tech world has gone virtual: Servers, applications, desktops, data centers, even jobs–at least to the tens of thousands of contractors doing full-time work at part-time pay with no benefits.

The one exception to the rule that virtual is better than physical is the one thing on which all the other advances in virtualization, cloud computing and job-creation avoidance depend: the network. At least, that’s according to John Chambers, CEO of the company with the most to lose from any wholesale separation of the value of a network from the underlying hardware that makes it possible.

High-quality, high-speed, reliable, secure, multi-functional networks require both good hardware and good software, Chambers told investors and analysts on Cisco’s earnings call this week (transcript via SeekingAlpha)…

It’s obvious that large-scale networking can’t be done at all, let alone be done well, without solid hardware underlying the software that controls all the action.

It’s only slightly less obvious–very slightly less–that Chambers wasn’t making a generically obvious statement of fact. He was making a policy statement.

Anything that would reduce the importance of the name painted on networking is an obvious Bad Thing in Cisco’s view.

Virtualization in this case means software-defined networking (SDN)–which is the TLA created to differentiate the flexible, dynamic reconfiguration of enterprise networks from virtual private networks that do little more than encrypt and decrypt network traffic between a client and server to make remote access more secure.

Chambers’ effort to reiterate the indivisible union of Cisco software and Cisco hardware in the networks of its customers comes just as the company seems to be (financially) recovering from strategic confusion resulting from too much focus on its own ambitions and too little attention to the needs of its customers.

Rather than focus on sophisticated networking, Cisco went all in on videoconferencing, VoIP, digital signage, consumer video cams and a dozen other things that did little or nothing to advance Cisco’s own core competency (which may be managing giant companies going through repeated mergers, rather than anything having to do with technology, but that’s a question for another time).

In April 2011, Chambers sent a memo to employees acknowledging that the overly diversified company had "lost the accountability that has been a hallmark of our ability to execute consistently for our customers and our shareholders," leading to a yearlong reorganization, reconfiguration and wholesale recast of its strategy.

The process was more of a reorientation than a reboot. Cisco is still focused on making the transport and management of video and voice, virtual meetings, virtual collaboration and digital signage as integral a part of its product set as the movement of packets.

It doesn’t see other virtuality as a critical element, even though SDN–or something like it–is widely accepted as the next logical step in the progression of flexible network infrastructures–one that matches the agility, performance and dynamic reassignment of resources provided by the cloud, virtual servers and all the other bits of virtualization that turned the IT industry on its head a few years ago.

SDN isn’t about remote access. Its goal is to make configuration and management of large networks simpler and more flexible by letting network managers define capacity, membership, security, usage policies and other functions using software that doesn’t care who made the hardware it’s reconfiguring.

SDN is still a nascent technology. Definitions are still flexible, and the percentage of adopters is still in single digits, so Cisco isn’t catastrophically hidebound on the issue yet. It was one of the first to build virtualization into its own products, but mainly virtualization of products from other companies, not from Cisco.

Cisco’s SDN strategy depends on its proprietary Open Network Environment (ONE), which–oddly for any type of virtualization plan–depends partially on features and functions built into Cisco routers and switches, but not anyone else’s.

Cisco also offers Overlay Transport Virtualization (OTV) and Virtual Device Context–proprietary network virtualization functions as Layer 2 networking extensions–to support bridged connections between data centers within a private cloud network.

It’s not exactly unique in that, though. It has moved slowly enough in the race to create efficient ways to connect multiple WANs (without gumming up Ethernet layouts or changing existing applications) that HP’s comparatively tiny networking group was able to catch up to, according to Network Computing.

Painting Vendor Lock-in With Happy Colors

Tying network hardware and software inextricably together may provide some demonstrable technical advantages–Cisco claims better security and app performance optimization by tying to its hardware–but requires users to stick with Cisco hardware or buy only from other vendors that are close allies of Cisco.

High-performing, reliable networks, Chambers implies, can work only with existing, proven networking technology; virtualization can be added as a programmable layer of software across the top. "That’s a tall order," according to the wryly dry understatement of Network Computing’s Mike Fratto.

Not only is it a tall order, but it’s also one that handicaps both it and its customers with the need to preserve Cisco’s legacy hardware business at the potential expense of customers trying to build networks that are flexible and configurable according to their priorities, not Cisco’s.

Of course, Cisco is the company that forgot to ask permission before forcibly upgrading the firmware on some products and replacing the on-machine management interface with one that forced users to create a new account on Cisco’s Connect Cloud in order to manage devices they already owned. It also warned it would be keeping track of "certain information related to your use of the Service, including but not limited to the status and health of your network and networked products," which is enough to give even the least paranoid network admins nightmares of Big Brother.

Cisco did back down and allow users to opt out of its supervised management interface. The whole thing smacked of arrogance and a powerful effort to ignore the potential impact on or reaction of customers–very much the kind of thing a company would do that hadn’t quite reformed its lack of attention to its own customers, as Chambers said Cisco has done.

Overtly stating that one’s hardware products would be inextricably linked to the virtualization software providing functions users want to add also smacks of arrogance–the kind that assumes Cisco customers should toe the line on its interpretation of virtual networks rather than their own needs or desires.

It’s not the kind of position you’d expect from a company that would willingly support a customer’s desire to cut the ties between control of a network and the provision of it.

I don’t know what will be the specific benefit that makes virtual enterprise networking take off. Judging from the behavior of companies big and small over the past five years, however, I’m pretty confident there will be a killer function or two and that virtualized big-time networking will take off, whether as SDN or some other acronym.

I’m also reasonably confident that Chambers and Cisco will either back down or add some flexibility to Chambers’ apparent belief that networking can’t happen without networking hardware–and that networking hardware isn’t worthwhile unless it’s from Cisco.

It’s a hard line, one that punishes customers for buying from Cisco and for having plans for their networks that may not mesh with behavior that will optimize Cisco’s financial future.

Chambers emotes warm fuzzies while promising customers that Cisco will respond to as great a degree as possible to their individual concerns. The fuzzies fly so thick it’s hard to credit them as sincere.

But, as in the Cisco Connect cloud fiasco, he does change his mind and, more importantly, change the behavior of Cisco as well.

It’s possible that, by the time the installed base of SDN grows from 2% to something formidable, Chambers and Cisco will be fully on board with the idea that hardware is simply hardware and that software is what makes it flexible, responsive and dynamic.

It’s also possible HP and Cisco will stick with proprietary, mutually exclusive approaches to SDN and get so pissy about the rivalry that neither is willing to open a crack that would make it easier for customers to breathe.

What’s certain is that, in the long run, hardware is simply hardware. Neither network managers nor end users pay Cisco to truck over yet another big box to sit in a data center, sucking up power and blowing off heat.

Customers pay IT vendors for new capabilities, new functions, for the flexibility to create, change or expand their own systems according to their own needs. They don’t pay for pretty boxes with Cisco’s logo painted on like a VoIP phone product placement in a prime-time drama.

Chambers and Cisco will figure that out quickly enough and figure out how to add more of the unique benefits of a combined hardware/software Cisco solution into the software alone.

They’ll figure it out, or customers will go elsewhere to buy software they can use to run networks without having to worry about where the hardware came from, whose name is on the side or what John Chambers thinks of the way their networks run.

That’s their business, thank you and, whenever possible, it’s one they’ll handle without being forced to dress a certain way, act differently or buy all their hardware from one place simply because the vendor’s philosophy and financial results depend more on vendor lock-in than customer satisfaction.