Solid Planning Assures Smooth Telecommunications Upgrade

• By Charlie Henderson
• 06/01/02

In 1999, the people in charge of information technology at the University of Minnesota (U of M) could see the end coming. The end of the useful life of the telephone system, that is. The university’s Twin Cities campus had grown beyond the ability of the original mid-1980s phone system to keep up. The phone system also lagged technologically in some ways, but that was to be expected after a decade and a half.

The Twin Cities campus, in Minneapolis-St. Paul, hosts more than 45,000 students. It is a geographically rambling campus that was served for 15 years by three separate Intecom S80 PBXs.

Getting an early start on the process, administrators extended a request for proposal to several telephone equipment manufacturers. It also solicited proposals from telephone companies to explore telco-provided solutions, even though the university had operated its own phone system since 1984.

After weighing five proposals -- three from equipment makers and two from telcos -- administrators concluded that the most cost-effective approach was an upgrade to Intecom’s PointSpan IP-enabled voice communications system. This strategy presented the most cost-effective reuse of the cable and distribution center infrastructure, as well as the simplest means of managing the change.

Final cost of the upgrade was pegged at about half of the $31-million cost of the original network installation. One factor that eased the upgrade cost was the far-sighted decision by university administrators in 1984 to install a core infrastructure of fiber-optic cable in addition to all-new twisted pair copper wire. Through the years, that infrastructure was well maintained, so no significant cabling costs were incurred in the upgrade.

Administrators had certain specific requirements -- driven by the need to provide uninterrupted and easily manageable telephone service -- for its new Twin Cities campus network:

- resolution of its capacity problem;

- decreased space and power requirements;

- simplified maintenance;

- improved service provisioning to customers; and

- more reliable, fault-tolerant telephone service.

The old hardware wasn’t fault-tolerant, and because the system had three main switching locations, this tended to create large outages when failures struck. The university also became incapable of handling massive call volumes in times of emergencies or special events. In those cases, customers had to be removed from service just to keep the entire system from failing.

Another must-have was a single database system. The old system was split into three databases, which meant that the U of M technical staff had to work in all three at the same time to make sure that services were built correctly and effectively synchronized. This complicated daily operations and added unnecessary labor expenses.

When the technical staff planned the project, they looked closely at how it would affect the university’s business, since the upgrade was going to be done in phases through the course of a school year. Because they stuck with the same system vendor, the station equipment required no end-user training, and the staff already had the necessary technical skills.

Administrators created a detailed plan for the project, and posted that plan on the U of M Website. It can be seen at . It spelled out the year-long process of the upgrade, from initial replacement of telephone sets in mid-2001 to the completion of the project in mid-2002. The St. Paul portion of the campus was the first on the new network, and then, building by building, the remainder of the 210-building campus was added.

The equipment rooms offer insight into the processes that helped ensure a smooth upgrade. First, technicians prepared the equipment room and installed the new core equipment, running both systems simultaneously until the cutover was successfully completed.

Administrators imposed a database “freeze” while database information was extracted from the old PBX and imported into the upgraded PointSpan equipment. Cutovers were timed for Friday nights and, at that time, the old PBX database was modified to route the user’s directory numbers to the upgraded system. Each phone was added to the new system by moving cross-connect wires in the equipment room. The old PBX cabinets were powered down and removed when all services were removed.

Once technicians were satisfied that the equipment and service were stable, operational control was transferred back to the Office of Information Technology and the database freeze was removed. Any additions implemented during the freeze were reconciled. Most sites were completed in a single weekend in this way.

One bonus of the system is a smaller footprint and lower power requirements. Several equipment rooms were practically bursting at the seams, and the old systems took up about 1,000 sq. ft. and were consuming more than 400 kilowatts per hour. Heat output was measured it at 1.24 million BTUs per hour.

The system upgrade made a tremendous difference in terms of space and power. The new system, at a total of about 460 sq. ft., is less than half the footprint of the old system. It consumes only 48 kilowatts per hour -- about one-eighth as much as the old system -- and it produces about 175,000 BTUs per hour, or only about 15 percent as much as the old system. This has drastically reduced cooling needs.

Clearly, the upgraded system is meeting U of M’s objectives.

- Capacity problems are resolved, with vastly increased directory number limits. The university can now handle half a million calls per hour, nearly 10 times more than it could before.

- The single database and point of entry eliminated the need to keep multiple systems synchronized, and the new system links with on-campus databases, such as the X.500 e-mail directory.

- Administrators implemented a common numbering system across the entire campus. Previously operating off three separate PBX systems, there were three different numbering zones.

- Strategic location of servers in different buildings assures a high degree of disaster recovery. Either of two main Enterprise Communication Servers can run the entire 40,000-port system, while six geographically separate Node Control Servers provide survivability in the event any node gets isolated from the network.

- Services can be offered, like last call return, enable/disable Caller ID presentation, distinctive ringing, call forwarding, selective call forwarding, anonymous caller rejection and call waiting caller ID. Also, the system enables enhanced 911, so caller ID information follows a call transfer.