Of course it’s not quite that simple. I’ve just decomissioned an old Red Hat 7.1 box (hosted dedicated server) that had been in service since 2002, so about 7 years. Specs? Celeron 1.3GHz, 512M, 60GB HD. Not too bad in the RAM and disk realm. It did a good job but goodness am I glad to be rid of it!
Not having that box online is safer for the planet, although it (perhaps amazingly considering the age of some of the externally facing software components) has never been compromised – I consider that mostly luck, by the way, I’m not naive about that. But it’s not easy to move off old servers, it’s generally (and also has been in this case) a lot of work.
Of course hosting has moved on since 2002, places like Linode offer more for less money/month. Of course they virtualise (Xen based in this case) and that’s not been my favourite (particularly for DB servers but depending on the use it really comes down to how you set up the whole infra). It’s a different environment, so different “rules” apply for the optimal setup. The feature/pricing model of the hosting(/cloud) provider actually has more than a little bit to do with that. Distributing tasks like MX relaying, DNS, moderate MySQL tasks, web server, across different virtual machines, with added redundancy across different data-centers, works very well for many use cases. And the funniest thing… more servers, with distributed redundancy, the net cost per month is actually lower than that one single server!
There a many aspects to consider, and I’m intending to write more about that in future posts. I just found it an interesting experience, dealing with this (personal, not even business) server. We handle with these technical environments all the time in our work, but it’s not quite the same perspective. It’s not all technical/financial issues, there’ more to it.