IT managers who must determine the cost to roll out wireless LANs (WLANs) throughout their organizations often focus on the cost of access points and WLAN controllers. However, there is much more to determining actual costs than these initial capital expenses. Depending on the type of WLAN equipment deployed and the intended applications, companies may be forced to overlay two sets of access points, upgrade client devices, add extra access points because of weak technology that limits capacity, and to spend additional money on pre-deployment studies as well as ongoing configuration. Before deploying a WLAN, enterprise should consider the variable costs that go beyond simple equipment list prices to ensure that the technology they are investing in maximizes productivity while minimizing cost. Pre-deployment: Measure Twice, Cut Once When deploying a wireless LAN, it is very important to define what the organization expects from the network. Some technologies are better suited for certain tasks than others. In a larger enterprise or campus situation, not understanding the target use for the network can quickly lead to iterative upgrades in short periods of time. Also, understanding how the network will be used upfront, versus a couple years down the line, will help lower the total cost of ownership as well. Common applications in use today are e-mail and web access with Wi-Fi deployed as a more convenient way to access them, but accessing increasingly advanced, mission-critical applications wirelessly is becoming more popular for providing increased productivity. Voice over IP (VoIP) has emerged recently as one such mission-critical application. According to Infonetics, 72% of companies will be using wireless VoIP by 2009 . This is not surprising, given that the growth of VoIP is outstripping traditional voice PBXs in enterprises, wireless VoIP handsets prices are dropping 15% per year , and the WLAN is a natural extension of the wired LAN that is already carrying VoIP. This means that even though VoIP may not be a priority for enterprises today, having an infrastructure that is ready for that application, while satisfying current, data application needs, is critical to avoiding forklift upgrades in the future. Understanding the applications also helps determine network design. For instance, if VoIP is planned for the future, the network may have to support a very high density of mobile clients. Adding VoIP to the WLAN means that some users will have two or more wireless clients each -- a wireless VoIP client and their laptop or PDA. This conservatively doubles the client density expected on the network and, therefore, dictates network design today, while avoiding labor-intensive redesigns in the future. Deployment: Right Tools for the Job Having scoped out the applications for the WLAN and the density of users that must be supported across the campus/enterprise, the next step is to design the network in the most cost-effective manner possible. To determine the cost of the deployment and how technology can help, it's important to understand the elements that comprise the total cost of deployment. Wireless networks are much more than the cost of an access point (AP) and the management equipment (switch or controller). Every access point requires an associated Power over Ethernet (PoE) port, a cable run to the place where the AP will reside, and the time associated with deploying all of these. An access point that costs $200 is actually more in the $500-600 range when the planning, labor, and PoE are taken into account. In fact, a Gartner study shows that the capital costs of a WLAN comprise as little as 15% of the cost of installing a WLAN when client upgrades and administration are factored in. This is where technology can help. Knowing the limitations of 802.11 and how advanced systems are designed to help can save on equipment as well as labor costs. For enterprises that require basic data connectivity now and into the future, and then want to provide strategic "hot spots" across the corporate campus, the solutions are simple and plentiful. Since advanced applications are not a concern and the wireless network is available simply as a convenience, the basic enterprise systems will suffice. Labor is reduced through centralized management and security, and there won't be many quality concerns with the Radio Frequency (RF) spectrum since Internet Protocol (IP) data is very tolerant of errors and the user density would be relatively low. If, however, the WLAN is seen as a strategic network initiative that will eventually carry VoIP and other advanced, mission-critical applications to employees, then more advanced technologies are needed to create the continuous, pervasive coverage that these applications require. Given the time sensitive nature of VoIP and the density of clients that it creates (users carrying multiple clients), the right technology can alleviate costs significantly. Microcell Design: Costly Mistake Many WLAN vendors believe that the only way to support high user density and applications that require certain service levels is to deploy more APs with smaller coverage areas such that each AP need only support a few users at a time. Commonly called "microcells", this practice of reducing AP output power to create smaller cells, add extra APs, and ultimately support higher data throughput, is a costly solution for enterprises. Having established that the cost of deploying a WLAN and its associated APs is optimistically 30% capital expenditure and 70% labor and infrastructure, this design clearly increases equipment and labor costs and still does not provide the optimal environment for advanced applications. The reality is that most WLAN access points were not designed to handle a pervasive deployment -- that is a deployment where coverage is continuous and available everywhere, not just in select hot spot areas. Trying to shoe-horn wireless technology that isn't designed for pervasive deployments creates some key problems for more advanced applications and leads to higher costs. The problems include: Reducing data rate vs. cell size With a Wi-Fi access point, the data rate reduces as you move farther away from it. So, to achieve higher overall throughput, access points must be placed closer together, i.e. microcells must be created. Limited non-overlapping channels The 802.11b and 802.11g wireless LAN standards operate in the 2.4 GHz band. This band only offers three non-overlapping channels. Each channel is a separate 'pipe' of bandwidth. All clients attached to the access point share that same bandwidth. Collisions and interference As wireless clients are not physically connected together, they are unaware of when others are trying to transmit. As a result, it is likely that multiple clients in the same area will transmit at the same time. When this occurs, the clients are programmed to 'back off', reducing the performance of the client. If the client can normally transmit at 54 Mbps, but has to wait due to other collisions, then the effective transmission rate is reduced because of the time spent waiting. Handoff between access points When a client roams from one access point to another, the time between disconnecting from the first access point and reconnecting to the second access point is non-zero. For some clients, this process can take up to several seconds. If a micro-cell configuration is deployed to increase throughput, then this handoff problem is exponentially worse, as depicted in Figure 1. Figure 1: In the same physical space, the client on the left experiences one handoff vs. six handoffs in the same area using a microcell deployment. Multiple handoffs for voice clients can significantly degrade quality. The cost of dealing with these problems is realized when IT staff must try to account for them through RF and channel planning during deployment. Using advanced software tools to manually predict the best placement of access points, so as to minimize these problems, is labor-intensive and ineffective when tens or hundreds of access points are being deployed. Enterprises need WLAN systems that are designed specifically for pervasive Wi-Fi access in enterprises; these systems go beyond simple centralized management of APs and security with software tools to help deployment, and delve into application-specific performance enhancements and automated deployment optimization. By handling the issues of pervasive WLAN deployment automatically and essentially eliminating them from the design process, these systems reduce deployment costs substantially. These systems require no RF planning, eliminate quality-killing handoffs, they can eliminate interference problems and enable greater scalability. Understanding the enterprise's requirements today and into the future will help identify which type of WLAN System is required. Either way, matching the network to the enterprise's needs will avoid costly upgrades in the future. Tuning, Reconfiguration, and Ongoing Management The biggest downside to having the wrong pervasive WLAN design is the ongoing tuning and reconfiguration of the network. If the WLAN system requires RF planning tools to deploy, this will be an ongoing task. The only thing constant about the nature of Radio Frequency and mobility is change. Even if it were possible to create a microcell design that solved all the potential problems and even if the enterprise is willing to absorb the increased cost of all the extra access points needed, there will still be changes in the network that create problems and require more adjustments. Moving a desk, adding employees, or even employees gathering around a water cooler in a way that was not originally predicted, will change the RF environment and, therefore, require changes to the network. If those changes are not handled by the system automatically, the IT manager must manually intervene with software tools. This process is commonly referred to as RF Spectrum Management and is the bane of IT staffers' lives. There are two ways to avoid this added labor cost -- First, the enterprise could use only web and e-mail traffic on the WLAN as these applications are very tolerant of poor performing networks and likely won't generate many support calls. Second, the enterprise could deploy a system designed to automatically deal with interference, dynamically assign quality of service to high-priority applications, eliminate handoffs between APs, and automatically adjust to changing conditions. Conclusion Understanding the costs that go into deploying a pervasive wireless LAN is the key to right-sizing the network to adequately address any enterprise's needs. These costs go way beyond the price quote a vendor delivers based on the size of the enterprise. Some research into the usage of the network today, and planned usage in the future, will allow IT managers to select a system that can support both current and future needs, while minimizing costs.