Virtualization and Software Cost

Chuck Hollis scares me. If you aren’t a regular reader of his blog you’re missing out. He has a deceptively disarming way of painting the picture of technoloy’s future that makes you feel better about the coming apocalypse. Chuck just put out a series of great posts on The Changing Face of Information.  He posts often on VMWare and virtualization in general and for the series he included this post where he argues that virtualization reduces friction in IT delivery.  It makes me wonder, with no friction – how do you hold on to it. Or better – how do you hold on to IT.

The Slippery SOAP

I won’t agrue the lubricating effect of virtualization. That point is fairly obvious. As we adjust to this new way of doing business, we have to remember that it is in fact business. More than the technology changes. Conceptually I have been struggling with software licensing models in a virtualized enterprise. Buying the capability that software gives you is abstract in the first place. Charging by named user becomes a confused proposition when services perform the most significant functions of a system. This idea first played out with N-Tier architecture and SOA makes it even less concrete.

Some time ago, IT organizations and vendors alike recognized the diminished relevence of named users and sought out a different quantitative value to use in pricing models. Along came CPU based pricing. Accelerating clock speeds made the proposition weak over time anyway but virtualization all but invalidates the model. You could argue this is just math but how many virtual CPU’s you operate determining what you pay for won’t play out in the real world. 

You Get What You Pay For?

Companies simply will not purchase based on the peak CPU draw that a given application might use. The customer will want optimize to the low end and haggling could go on for years. A good deal of retooling or blind faith would have to be built into maintenance agreements. This type of disruption to revenue models of vendors will not be sorted out over night, or over a weekend or even over a budget cycle. (and I didn’t even mention SaaS- that’s for another day)

Software vendors have to explore and innovate new ways of monetizing the value their software provides. The Web itself has been churning out new business models for years and it seems that churn has finally washed over enterprise software, erasing the lines in the sand we had drawn to define what we were getting for the investment.

The Path to the Dark Side

I have to be careful that my years in IT don’t overly bias my opinion but no friction is potentially as bad as too much. As Chuck stresses, the way we think about IT infrastructure needs to change.  Unfortunately, the temptation to follow he path of least resistance can have disasterous results. If you remove the constraints of physical resources and don’t step up the discpline around management of the allocation, that new found capasity will quickly be swallowed up by poor performing applications.

I can’t count the number of times someone immediately suggested throwing more hardware at a performance problem but when that proved too expensive, they went in and fixed the query causing the problem.  When adding capasity is easier AND cheaper than changing code – it’s the end of the world as we know it. I seriously don’t want to depend draconian operational practices. I LOATH them. These are however some of the harsh realities we have to think about enterprise software value and control in Chuck’s brave new world.

Comments are closed.

Create a website or blog at

Up ↑