The Campus Untethered

Fancy computer labs and state-of-the art libraries are fine, but when the mouse hits the pad, “Today’s students expect wireless access in the classroom, in the common areas and in residence halls,” says Albert Stadler, network administrator for Missouri Southern State University (MSSU).

At MSSU, with about 5,500 students and 450 faculty members, it’s an expectation that’s taken quite seriously. Over the last year, the 365-acre campus in Joplin, Mo., has installed wireless access in 16 of its 23 core buildings, providing secure connectivity in classrooms, dorms, the cafeteria and the library. “Wireless technology is helping the school move into the 21st century,” Stadler says. “We’re cutting costs and boosting efficiency.”

MSSU enjoyed major capital savings by not having to rewire any of the dormitories to increase capacity or functionality, says Stadler. Also, students appreciated the convenience of being able to work wirelessly in common and outdoor areas rather than be confined to their dorm rooms.

Like MSSU, a growing number of universities are cutting their connections to legacy Ethernet and migrating toward wireless campuses. “Universities are among the early adopters of wireless LANs [local area networks],” says Rachna Ahlawat, principal analyst at Gartner Inc., a Stamford, Conn., market research and IT consulting firm. She notes that about 25 percent of wireless installations are now at universities.

Unlike users in the typical residence or business, users on college campuses want access both inside buildings and outside. Furthermore, a variety of users—including students, faculty, administrators and visitors—with different security levels and privileges require access. That makes “authentication, encryption and security key considerations,” Ahlawat says. Finally, there are technical issues to tackle, including managing bandwidth and ensuring quality of service (QoS). “It is important to address all the challenges,” she adds.

Schools that put all the pieces of the wireless puzzle together effectively can reap significant returns. Not only can the campus gain stature as a tech leader and establish itself as an attractive destination for students and faculty, it can usher in more effective learning and reduce time and money spent on wiring.

Cutting the Cord

At Missouri Southern, Stadler realized early on that the school had to go with the wireless flow. After successfully deploying the technology in a couple of test areas, the school opted to roll out service across the campus. However, the proprietary nature of the existing systems created immediate limitations—especially with an IT network staff of only two.

The university relied on medium access control (MAC), address-based authentication and security methods, and it could not provide the tracking and bandwidth control required to manage a diverse group of users. The tracking ability was particularly important so that the university had a record of which device used a particular line access at a given time and to notify students of any viruses on their computers.

Stadler moved to a more scalable environment built on advanced routers and switches, as well as an 802.1x authentication system. The latter provided a more robust infrastructure that could handle tracking, user IDs and MAC address management. The university is now in the process of installing nearly 100 additional access points (APs). Students and faculty can use a single user name and password for the wireless and wired networks. At present, about 30 students per hour authenticate.

The system has reduced confusion and support costs, says Stadler, who expects there will be fewer help desk calls and less manpower invested in troubleshooting wiring problems during the first full semester of use.

Gartner’s Ahlawat says that it’s best to map out a wireless environment before going live. Design elements—including how the school lays out all the access points—play a key role in determining overall performance and cost. As a rule, each AP has a range of about 800 feet in an open environment. Rather than setting up a wireless network using a trial-and-error approach, it’s wise to conduct an upfront analysis and base decisions on real-world data.

Ultimately, it’s important to provide services tailored to the required load. Simply tossing APs and other technology at the problem can run up costs. Scaling to the maximum load can increase the cost for each location as much as fivefold, Ahlawat notes. On the other hand, inadequate scaling can result in three or four students not being able to connect to the network through any given AP. “An instructor might have 30 people in a classroom, and if a few of them cannot gain network access, it can undermine the entire process,” Ahlawat says.

That’s a point that Thomas Skill, CIO and associate provost at the University of Dayton (UD), understands well. In 2000, the school— which has 58 buildings on a 200-acre campus in downtown Dayton, Ohio—began rolling out an 802.11b wireless network in classrooms, the student union, library, commons and student housing. He knew that the system had to connect students and faculty quickly and simply. “Our biggest fear was that users—particularly faculty—would reject the technology if it didn’t perform well,” he says.

Among other things, that meant mapping out APs, provisioning bandwidth and installing authentication so that there was zero margin for error. His engineering staff conducted site surveys and then began installing APs. The school now has 653 in place. It uses sophisticated system logging that provides data about the number of unique connections and data throughput.

“The data is extremely enlightening,” Skill says. “At different times of the year usage patterns change, and we can provision the network appropriately. It has also helped us understand how and where to expand the network.”

With APs, “it’s important to establish your goals and understand your user demands upfront when tackling projects involving wireless,” says Skill. “UD does not deploy the same platform of AP everywhere. We evaluate how many users are going to be associated with an access point, the type of traffic going through it, the surrounding environment and how fault-tolerant the design needs to be.

“Achieving a good deal of fault-tolerance is not something that you hear a great deal about with wireless,” Skill says, “but when dealing with areas such as classrooms, it becomes very important to provide additional coverage and load balancing via multiple access points for critical ‘hot spots.’”

So far, the university has spent about $400 per AP on campus and $100 for those used in the residential areas. Every wireless-equipped classroom has at least two APs. This summer, the university also invested over $400,000 upgrading the network that supports the 350 homes the university owns. Says Skill: “Students expect to have access to a wireless network. It is part of their basic mentality. Our goal is to put the systems in place that make that possible. We know that it’s used for everything from classroom instruction to surfing the Web.”

Track Down Illicit APs

One university that is building a better wireless network is the University of Texas at Austin. The 50,000-student campus began deploying a wireless LAN in 2000. At the time, it had fewer than three dozen APs. Today, it has deployed more than 1,200 APs and is on its way to installing nearly 2,500.

All the school’s libraries, the student union and many commons are equipped with wireless. Approximately half of the classrooms are also equipped for wireless access. About 87 percent of the areas use 802.11g, and only a scattering of 802.11b and 802.11a exists. Usage has grown by 72 percent over the last year.

“Manageability of such a large number of access points is a difficult task,” says Dan Updegrove, the university’s vice president for IT. The University of Texas at Austin uses a combination of vendor and self-designed tools to monitor the network.

“For the first couple of hundred access points we could throw people at it and configure a lot of things by hand,” says Keith Nelson, director of telecommunications and networking. “Once we broke through the 500 barrier, we knew that we needed to have everything automated.”

“We started deployment of the wireless network in 2000,” adds William Green, the assistant director of the school’s informational technical services department. “The way things work at this university, things tend to be funded by departments in a sort of demand-and-pull model.

“We don’t come up with a large sum of money and roll things out. We allow consumer demand to drive where technology is deployed. We started with about 20 to 30 units at popular hangouts for students on campus. Today, we are at about 1,200 access points. We expect to eventually wind up with 2,000 to 2,500 across campus—if we decide to go for completely ubiquitous coverage.”

Currently, the school uses software that autodiscovers and autoconfigures APs and blocks deployment of rogue and personal APs. A particularly vexing problem with rogue APs is when students or faculty encounter problems with one, they consider those problems as being the fault of the university’s wireless network, not a rogue AP. Another problem with rogue APs is they’re usually configured incorrectly, which can cause interference problems with university APs.

Fortunately, the university’s system can track down illicit APs. The school handles quality-of-service issues and inhibits the use of peer-to-peer applications by allocating a strict quota of bandwidth to users. A user who exceeds his or her quota must purchase more. For security reasons, the university maintains separate wired and wireless networks. “There are areas where we need to ensure reliability, performance and security,” Updegrove says. “We’re not ready to run the entire system under a wireless cloud.”