January 23, 2004, 3:53 PM — While security remains at the top of IT's list of issues with wireless LANs, I think we'll soon see attention shift to what really is a much more significant question (more significant than security? Indeed!) - and that's the question of total cost of ownership, or TCO. I've found just about every other issue pales when money becomes the topic of conversation, and it's really a positive trend that we're now getting lots of questions about how to cost-justify a WLAN. Few enterprises - less than 10% - today use WLANs on a mission-critical basis in general-office applications. That's going to change dramatically this year, as prices have fallen, performance has improved, security (yes!) has been addressed, and clients become almost free. But how do we go about evaluating the TCO of WLANs?
The first step is to develop a model of cost of ownership. This involves breaking cost down into two major components: capital expense, or CAPEX), and operational expense (or OPEX). CAPEX includes the cost of any equipment involved, and OPEX includes installation, maintenance, and ongoing operations. Over time, OPEX is much more important than CAPEX, since CAPEX is a function of the always-declining costs of hardware and software, while OPEX usually reflects labor costs, and these only grow over time. Nonetheless, both can be favorably addressed, and the TCO of a WLAN infrastructure successfully managed and justified.
As it turns out, both elements of TCO can be highly influenced and favorably addressed primarily by a choice of WLAN system architecture. As I've discussed before, centralized (often called "switched") architectures have become popular over the past year primarily based on a CAPEX argument - since the "thin" access points used in centralized systems cost less than those used in traditional distributed configurations, the additional cost of the wireless switch is quickly amortized and thus the entire installation should cost less. With emerging standards, which will lead to commoditization, and still-declining component costs, thin APs will eventually be almost as cheap as clients - and they're almost free as it is.
But, as it turns out, the high degree of product integration found in most centralized implementations can have a profound and beneficial impact on OPEX as well. Security, for example, needs to be completely implemented in the wireless LAN system chosen, including VPN termination. Ditto for subnet mobility and VLANs. In fact, we can sum up the best approach to minimizing TCO in WLANs as one of implementing the WLAN as an overlay that has minimal - if any - requirement for modification to the existing wired network. That's the best way to hold down OPEX, which, as we noted above, tends to dominate costs over time. The choice of specific WLAN products must be at least moderately influenced by its implementation of features that hold down operational expense.