VMware "Cost Per Application" – A Swing, and A Miss…

The aim of this post is not to rubbish the guys who put together VMware’s recently announced “Cost Per Application Calculator”. I can only imagine the amount of effort that would have gone into producing such a tool. But although a good job has been done I can’t imagine why anyone would find it useful, irrespective of how accurate or inaccurate it may be.

There are a number of issues I have with this kind of thing, but first and foremost is that does anybody actually think acquisition cost really matters? Because in the grand scheme of TCO, acquisition cost is pretty much insignificant. It may cost a few grand for a 2 socket box, but it will cost 20 times that or more to support and run it over the lifespan of the box. Even bigger picture, how about that new datacenter you didn’t have to build / move into because you virtualised everything and chopped your existing rackspace in half.

Second to that, things like this do little more than provide another target for pigeons like Simon Crosby. Why call him a pigeon? He obviously doesn’t have the brain of one. But like the “rat with wings”, he tends to fly in, shit all over everything, then fly off.

No matter how accurate you could possibly make such a tool, people will always pick holes in it. In my 100 application sample (input was 100 / Server A / Enterprise / Virtual / Average / Average), I got back an “8% cheaper with VMware” result. But what prices are they based on – list? What if I get my software from Microsoft 10% cheaper than what I get it from VMware? If the calculator assumes memory overcommit, how can it do so reasonably without know the kind of apps I have, which may influence the amount of overcommit I may or may not get? Why does the difference in SAN costs only reflect the difference in number of HBA’s, when given my sample (24 HBA’s for ESX / 36 HBA’s for Hyper-V) i probably had to buy another SAN switch as well? What if I was using blades, in which case I may have had to buy another chassis? Ad nauseam.

I know the calculator isn’t meant to be all encompassing, but if there’s no way that anyone in their right mind would take numbers spat out from a Vendor provided calculator and base some kind of meaningful cost model on them.

But the calculator does provide some value, in Appendix A of the results – the detailed breakdown.

Now in here, is some very valuable information for I think a lot of people either side of the thin line of Architecture. It’s not the numbers, it’s the categories. In my experience, there are very few people who see the big picture from end to end outside of Architecture teams. And this is fine – it’s not a reflection of anything other than the fact that in large enterprises, people generally need to be silo’ed in order to do their jobs effectively. Except Architects. So what the detailed breakdown shows is a kind of bigger picture that many admins and management might not ordinarily think about.

And to that end, providing some kind of prescriptive guidance on what people should consider when trying to figure out an acquisition cost would be much more valuable, rather than an online calculator. If I want to calculate financials, I’ll use numbers from my actual environment. But in the case where I don’t have a talented and friendly architecture team (*ahem*) to turn to for some guidance, then a nice document indicating the kinds of costs I should consider would be most useful indeed!



One Response to “VMware "Cost Per Application" – A Swing, and A Miss…”

  1. Duncan Says:

    I don’t think the calculator is targeted at Enterprise environments like yours Stu.

    LOL @ the simon crosby part

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: