These things are awful!
Fortunately, this isn't a quote from my users as I continue to roll out thin clients. It is a quote from a fellow teacher at another school where thin clients have been in use for some time. "Have you tried doing teacher training on these things? Ever tried running Geometer's Sketchpad? Ever tried running a class in a lab of thin clients?!!!" I won't print the words that followed, but he had nothing nice to say about the thin clients at his school.
I'm choosing to believe, however, that I've avoided a few pitfalls that plague the implementation to which he was referring. First of all, the expectations of staff and students at his school were not set for the real purpose of thin clients. I wouldn't expect Geometer's Sketchpad (or any other graphically- or processor-intensive applications) to run well on a thin client. These machines are, by nature, designed for basic productivity and web access. Want to run Office? No problem. Need to surf the web? Make sure you have a big pipe into your terminal server and you're golden. Running CAD? Not gonna happen.
He also had little good to say about their stability or reliability, especially when a full lab of the thin clients was in use. Screen refreshes become choppy, connections get dropped, and downloads crawl. Here, of course, are the bandwidth caveats that have continued to crop up in these posts about thin clients. Without adequate infrastructure, these things just won't give the performance they should.
When I first began considering the use of thin clients, I set up a test server on a single P4 machine with half a gig of RAM. I turned students loose banging away on it and definitely hit a performance wall at around 5 or 6 clients. So when I actually purchased the terminal servers I seriously overpowered them. I have finally found a use for 64-bit operating systems and it ain't on a desktop. These servers scream, though, and move around 64-bit chunks like they're going out of style. Further discussion with my fellow teacher made it clear that an overworked and underfunded IT guy (sounds familiar) tried to make a couple of aging servers stretch some very limited resources and avoid buying real computers to meet the needs of many students and staff. Great idea and the sort of thing we all do in desperation. However, as always, what we really need is real funding to make these projects work, whether we want to roll out thin clients, workstations, laptop carts, or 1:1 computing solutions.
As usual, it comes down to money. Fortunately for me, my super's feet were held to the fire and we had to throw some serious money at our recent tech refresh. If the money hadn't been there to do it right (and hopefully reduce long-term costs significantly), the thin clients would not have been my solution of choice. I probably would have been building a whole lot of cheap whiteboxes or swinging as many barebones machines from a first-tier vendor at educational pricing as I could. I'm afraid we all need Jerry McGuire on our side - someone needs to show us the money if we want to provide students and teachers with competitive, real-world technology.