In this issue, Microsoft senior network engineer Don Berry continues to recount the story of his company's experiences as an early adopter of wireless LANs.The Microsoft deployment of Cisco Aironet WLANs began three years ago with a couple of pilot tests. The first test involved a single building with 40 APs and 150 clients."Initially, we had a couple of design goals: high throughput and minimum required performance levels equivalent to DSL connectivity, or about 600K to 700K bit\/sec per user," recalls Berry. "So we needed to make the density [the access point (AP)-to-user ratio] support that."He says that translated into about 10 users per AP."We did lots of antenna, coverage and AP-mounting testing. We learned to space APs about 60 feet apart."How did Microsoft determine optimum AP placement on such a wide scale?"I got a printout of each building, took out my compass and drew circles" to figure out where the 60-foot coverage radiuses should touch, he says. He adds that in its manual site survey of the first building, the company learned that the APs couldn't transmit around windowed corners and he took that into consideration. But once Microsoft identified such variables, it created a deployment template it could reuse, rather than manually site-surveying each location.Berry said he conducted the site survey in March 2000 and had deployed 2,000 APs by October. "At our peak, we were deploying 200 APs per week," he says.Microsoft's biggest challenge was rogue APs, Berry said."Rogues are often set on highest power level. So they tend to create interference. And clients will [inadvertently] associate with them," Berry says.Berry says having someone with RF experience on staff is important. "In a wired network, you can troubleshoot right to a port or point of access because wires only run in one direction: lengthwise," he explains. "Wireless networks, though, embrace both width and height and require a multidimensional mindset."What does Berry think of the centralized management, site survey and configuration tools emerging from traditional and start-up companies?"They'll help the industry. We did spend a lot of engineering time" configuring and deploying the network, he says."However, I don't have a high confidence level of the management tools out there just yet. There are lots of good ideas, but their usefulness hasn't been evaluated and proven," Berry asserts.