Clients: fat vs. thin, net vs. note
My budget administrator has made it very clear that I need to have all of my ducks in a row very early for the FY 2011 budget process. The nearly year-long delay that municipalities tend to feel in terms of budget crunches is certainly going to be on her mind as she's reviewing my proposals for the coming year and as they're vetted by the school committee and superintendent. Stimulus money aside, there won't be any sneaking in items under the radar this coming year.
At the same time, though, I have a line item for leased equipment at the high school that I know will be re-funded in FY11 to avoid some very empty computer labs next fall. You see, we're due for a refresh in July that will cover all of the servers there, backbone switches, thin clients, and quite a few standalone PCs that will come to the end of their leases in the spring. To complicate matters, I have purchased a fair number of standalone PCs outright during the last 3 years to meet some immediate needs that I would ideally redeploy around the district to get all equipment at the high school on a single lease cycle.
The question is, have prices dropped enough for me to cover all of the computing needs in the school for the price of a lease designed to handle two thirds of the clients I currently have in the building? More specifically, the lease originally covered 5 servers (a storage server, 2 domain controllers, and 2 terminal servers), 75 thin clients, and 25 standalone PCs. In the interim, we've picked up another 40 standalone PCs and 10 laptops. It's OK, I'll do the math for you: that's 150 clients and 5 servers that I'd like to squeeze into a budget meant for 100 clients and 5 servers.
Thin clients might be the ticket here. I've made enough infrastructure upgrades that I think I could serve up 125 thin client machines over our network. I'd certainly be more aggressive in the terminal servers I ordered though. RAM through the roof, hard drive space galore, and a very serious look at virtualization technologies (both at the server level as well as to the desktop) rather than vanilla Terminal Services. The other 25 machines would still need to be relatively high-powered standalone PCs for graphics and processor-intensive needs that just can't be served well by a large thin client deployment.
On the other hand, commodity PCs are just plain cheap. I'm not talking the $299 Walmart specials, but it isn't hard to find solid desktop PCs on state bid lists for $5-600, monitors included, running Windows 7 Professional. Windows 7 and Ubuntu 9.10 both promise a lot of speed at very reasonable hardware specs and modern imaging utilities make maintenance pretty easy. I eliminate the single point of administration, but also eliminate that single point of failure.
Then there are netbooks. Oh so cheap and oh so handy. Not for everyone, to be sure, but certainly a possible alternative in some lab settings that might take a chunk out of my bottom line.
The instructors who regularly teach out of our thin client labs have expressed interest in standalone PCs as they seek to expand their curricula beyond productivity applications. Yet they haven't seen what 3 years has done to the state of the art in desktop virtualization. I don't have all the answers yet and I'll be talking with students, staff, and bean counters alike as I pull together my first draft budget in November. However, all options are on the table for improving student and staff computing while reducing (or at least maintaining) costs.