Timely, considering Rackspace’s recent IPO
We are currently in the biggest data center construction boom in history. At the same time, this boom is dramatically weakening the future flexibility and financial performance of information technology.
The quoted numbers don’t really tell you whether the incremental five million servers by 2010 includes upgrades for those already online, and I suspect some amount of replacement was buried in the whitepaper (or PR release) that was used as the source. Nonetheless, it’s a lot of servers that will eventually become obsolete, meaning hardware manufacturers must be chomping at the bit.
The fact that servers are a small part of the expense of building a data center can at once be spun as a deterrent to building the mini-me version for your own business and a sign that a lot of what gets built around the data crunchers might be overkill when it comes to maintaining 99.99% uptime. The decision is more complex that a mere cost of capital calculation, but by how much?
And what changes (if anything) about the full-loaded cost structure when the technology is proprietary?