Friday's post about my "datacenter fantasy" yielded a lot of great reader comments, ranging from "Duh, why don't you just do this in the cloud" to "What a great idea!" to "This might happen when hell freezes over." I'm paraphrasing, of course, but I thought it was worth some further discussion as I give really serious thought to making IT partnerships work in my town and cutting costs for schools, the municipality, and local businesses.
Given that I'm about as big a cloud supporter as there is to find in education, why didn't I suggest the cloud idea? It makes sense that we could very cheaply provision virtual servers and even desktops with the right cloud provider. This would probably be a worthwhile long-term vision. However, the constituents associated with this effort all have some investment in server hardware and software. Starting by simply moving the existing hardware to a secure, high-bandwidth location would save money and avoid too large a paradigm shift at the beginning of the effort. The real goal was to get the servers more highly available and out of closets or out from underneath desks.
Other readers felt that the goal was too lofty and certainly too expensive. I did call this a fantasy, didn't I? Even at our bargain-basement real estate prices around here, just the acquisition cost for the property would be significant. I don't doubt that substantial electrical upgrades would be necessary, but fire suppression and HVAC could probably wait. Most of the vacant buildings are air conditioned and this would not be a rack-dense environment. Keep in mind that we'd be starting with a few racks and several of the sorts of random towers that offices tend to accumulate when they lack IT and infrastructure support for planning server deployments. Cooling needs could be handled with standard office AC to start.
Another reader asked me via Twitter who would pay for the data lines. I envision a non-profit corporation or cooperative into which all of the constituents could buy who would manage and pay for the fiber and data connections (and the building and utilities, for that matter). Potentially, though, we could sell municipal bonds or find some other public/private funding mechanism.
Another reader had a great suggestion for getting started. Fiber, though inexpensive by the foot, remains quite costly to run and light. Copper, on the other hand, is incredibly cheap, as is the end-point hardware, and the bandwidth is easy to aggregate and apportion with simple devices. A copper-based infrastructure, while somewhat dated, could still provide substantial improvements for our area and help us achieve this fantasy (at least in some form to start out) at a lower cost.
I called it a fantasy, but if enough players could buy in, then economies of scale would allow for cost savings and long-term growth into a model that supported cloud computing, local high-speed access, on-premise computing, etc.