As I’ve said before, contrary to popular belief, there is more to fiscal responsibility than just going for the lowest price at all costs.
Recently a friend was working on a high availability solution for a small office that involved 2 VMware servers, a high performance switch, redundant Internet connections, and a SAN.
Now first of all, I can see we haven’t hit “high availability” since we still have a few single points of failure. They only have one switch (which didn’t even have redundant power supplies), and they only have one SAN.
Still, 2 servers are better than one, and with vSphere HA, vCenter, and vMotion, you can move VMs around, and even lose a host without impacting the end users.
That is, if you do everything that makes sense. In my networking class I made the analogy to my students between having a single $500 white box server, and building a high availability infrastructure for tens of thousands of dollars. The cost rises quickly, so you have to make sure have good reason to chase no down time. A static site that says “I have a cat” is not a good reason. Having a transactional site where an hour outage costs you $10,000, or where government regulation requires 100% availability is a good reason.
Once you’ve made the decision, then go ahead and spend the money. If you aren’t going to spend what you have to, then don’t go halfway. Save yourself the money until you can do it right.
So, back to my friend. For various reasons, an inexpensive SAN was selected that was not up to snuff for the task at hand. This made their migration extremely painful, took a massive amount of project hours, and actually reduced productivity. This was after the expense of good VMware tools, and enterprise grade hosts. Right now everything is back to local storage until the correct SAN can be purchased – if the client can be convinced to do so.
So now I’m coming back to the decision. Which do you prefer? Do you spend $35k and disrupt your office for a few weeks, or spend $50-60k for a flawless transition that is also (in the near term) future proof ? If you’re even considering the $35k, then you need the $50k. If you have no reason for this kind of system, then don’t do it. You can have a nice small office system for under $10k. But if you need more – then do it right or just hold off until you can. There is nothing worse than spending that $35k and being unhappy with what you ended up with…