Nearpoints

Craig Mathias

Previous Article Next Article

It's the OpEx, Stupid (and a couple of notable events)

Buying a WLAN is one thing; owning one is quite another

By Craig Mathias on Fri, 10/16/09 - 9:39am.

As part of my presentation on the .11n road show that I'm now in the middle of, I'm talking about a key theme motivating advances in WLAN (and really, overall network) management - that an emphasis on what it costs to purchase and install a wireless LAN detracts from where the focus should really be, and that is on total cost of ownership (TCO).

TCO includes both the cost to acquire the system and any non-recurring planning and engineering involved (this is capital expense, or CapEx), and the cost to actually use the installed system - and this is operating expense, or OpEx. OpEx can include all manner of seemingly trivial elements - service, support, maintenance, troubleshooting, help desk, training, backups, updates, and all of the hundreds of activities that go into day-to-day network operations. The problem is that, over the useful life of a given installation, OpEx will almost always exceed CapEx, with the gap increasing with the size of the installation. The reason for this is simple - capital equipment is on an engineering- and manufacturing-driven cost curve, with VLSI, manufacturing economies of scale and other efficiencies, and faster/better/cheap all working continually to lower cost. Couple this with the discounting prevalent in enterprise-class WLAN system sales, and this stuff is astonishingly cost-effective to buy if not just plain cheap considering what one gets today in terms of performance and features. But operations, on the other hand, is labor intensive. And while manufacturing costs continually decrease, labor costs only rise. One can counter this with improvements in productivity, however, and the best way to do this (much better than beating the operations staff to death, an all-to-common alternative) is with more technology, and a little planning.

The technology comes in the form of advances in WLAN network management tools. This is almost always the place I start today when evaluating a system or planning one for an enterprise. All of the vendors have .11n, so that's a given. Differences in architecture and implementation can make a big difference in performance, but it's difficult to evaluate these without a serious benchmarking exercise, and those can be difficult and expensive to do. But management tools are easy to evaluate, especially for network operations professionals, and I always recommend that these folks be part of the selection process for any new WLAN system purchase. As effective management tools augment the productivity of these guys, OpEx and thus TCO should be minimized, and the operations guys might even get a Saturday or Sunday off now and then. Productivity, indeed.

Two other events of note occurred this week. The first was the announcement by the Wi-Fi Alliance of the long-awaited Wi-Fi Direct specification, which pushes Wi-Fi technology into the machine-to-machine space, among other intriguing possibilities. Think kind of a cross between traditional peer-to-peer and soft AP. I know some of you were skeptical when I wrote that I thought Wi-Fi would have a major impact on personal-area networks, sensor applications, and even RFID over time. But I'm not kidding here - one world, a zillion apps, and one radio (OK, two if you count LTE). Atheros and Ozmo Devices are already on board with Wi-Fi Direct, and expect this to become a very big deal indeed.

The other was an announcement of an alliance between Brocade and Motorola that really is worthy of being called such. This is a two-way exchange of value that will yield new products and capabilities for both parties, and significantly gets Motorola into the wired networking business. I've been pushing the theme of unified networks for some time, with wired and wireless as cooperative elements, rather than wireless as overlay alone. There's also a cloud element here that is very intriguing indeed; more on that shortly.