As a small 3,000-student liberal arts school in Middletown, CT, Wesleyan University faces tough competition from much larger schools in the area. But with an infrastructure designed to grow and adapt to the school's changing needs--and a supportive administration--John Meerts, director of ITS, believes Wesleyan can get an edge on those schools. An "extensible" five-year plan is part of his vision
UB: Like many schools these days, Wesleyan's IT network has to serve a variety of purposes, from administrative to teaching environments. How is your network set up?
MEERTS: We're no longer dependent on mainframes, as we once were. We still run a VAX that's about 7 or 8 years old for our main financial applications. It used to run just about everything at this university, but now that's about the only application it runs. Most of our applications have migrated to Unix or NT-type servers. There are about 50 of these types of servers around the school
UB: Is that arrangement difficult to manage? Are there issues connected to the larger number of servers?
MEERTS: Originally, the strategy was to go to more independent boxes, precisely because if a part of that environment goes down, your entire environment won't go down. If you have 40 boxes, they are unlikely to all go down at the same time. That was the problem with mainframes--if one application brought down the machine, then entire system went down.
But, while that's been a good solution for us, I'm still a little nervous about the fact that we have so many different servers that are linked through software. Even if you have 40 servers that are up 99.9 percent of the time, that other one-tenth of a percent--times 40--gives you a sense of the failure that's possible within that complex. When the servers are tied to each other, as they are through software, you are increasing the risk of downtime.
UB: But as the technology needs of the school grow, don't the problems of maintaining so many servers increase as well?
MEERTS: Right. That's a real concern for us. Initially, it made sense to have multiple servers--or at least every substantial application had its own server--and that gives you some protection against that. But now we are worrying that that may not be scalable.
At what point do you start losing efficiencies because you have to maintain these servers independently of each other? Operating systems upgrades become more complex, or at least more voluminous as you have to do more of them; you have to do more database upgrades, and so on.
UB: What solutions are you exploring?
MEERTS: I think a possible solution is in some of these larger servers boxes, such as those from Sun Microsystems, that can be partitioned. It's a bit like the IBM VM (Virtual Machine) environment that was prevalent in the mainframe industry about 10 years ago. We are studying those issues and we want to be sure that, as we make our moves, we are aware of the potential consequences of those moves.
UB: Let's look at those investments in technology: How is technology spending planned at Wesleyan? Is there, for example, a focus on increasing the hardware budget?
MEERTS: Hardware is only a minor part of the budget. Our administrative computing environment has just created a 10-year plan that anticipates $26 million in expenditures over that period. Part of that is hardware, of course, but in the larger scheme of things, it's a relatively minor part. We're talking about maybe 10 percent of that budget being targeted specifically to hardware. The interesting thing is that our hardware costs have actually gotten lower, in part because of our low-end server strategy (most of our servers cost less than $50,000). With smaller servers, you can add as you need and it seems overall to be a cheaper solution. But we want to test that proposition as well--is it really a cheaper solution when you have to have the personnel to maintain 40 servers rather than four large servers? …