Scott Drummonds on Virtualization

Economic Theory and IT


[Note from Scott: Before embarking on this topic, I want to make clear that I am advocating no political system. I am using an cold war era economic analogy in support of my growing interest in improving the efficiency of IT. If you have strong opinions on the inherent Good or Evil of the political systems mentioned below, I politely request you air those opinions in a different forum.]

Several years I read an interesting book on the Cold War that documented some of the insane behavior of the superpowers that nearly culminated in the annihilation of mankind.  The book was rich in stories of espionage and assassination and political machination.  In one paragraph, buried somewhere in the book’s meaty center, the author includes an almost throwaway reference that has stuck in my memory ever since.

The story describes the USSR’s mining of titanium, which was used to produce a variety of products.  With no free market, the USSR lacked a proper valuation for the metal and the human energy required to process it.  Titanium is abundant in the region and there was a great demand for shovels.  With so much titanium, and no clear understanding of the value of the material and the cost of processing it, the USSR produced a lot of titanium shovels.

Without a proper value assigned to the product, the titanium shovels were sold and traded to the country’s neighbors at the going rate for “normal” shovels.  When the rest of the capitalist-driven world discovered the incredible bargain of titanium shovels and steel shovel prices, they bought USSR’s shovels at rock-bottom prices and converted the metal into high value items like rockets and planes.

The net effect was a tremendous waste of limited resources.  The valuable metal and limited human capital were directed into a venture that was not profitable and only beneficial to the USSR’s neighbors.  In addition to the story of shovels, the book related tales of unneeded refrigerators that piled up by the hundreds of thousands in warehouses and workers that dug canals through solid stone with wooden tools. These economic inefficiencies were among the many causes of the country’s collapse.

Here is where I get to IT.  If you are today running your virtual infrastructure with no clear understanding of the costs of your business with respect to value of your products then you are doomed to the same slow failure.  Your product–in this case virtual machines–have a value on the free market and will be bought, traded or wasted based on the cost you set.  Produce a “titanium” virtual machine for free and I promise you your customers will order it and waste it.  Produce a steel virtual machine at titanium costs and your customers will go elsewhere for their product.

To make your virtual infrastructure competitive in the free market you have to do two things:

  • Establish a currency that can be used by everyone in the industry.
  • Properly map your products to their costs and value.

Obviously what we are talking about here is a realistic chargeback system.  I know that many of you roll your eyes and shake your heads each time I talk about this.  But you have to realize that the free market is coming to you whether you like it or not.  Your internal customers are today being courted by public cloud service providers and if you cannot offer a better product at a competitive price, you are going to lose those customers.

Today you can survive without properly valuing your product (virtual machines) because you run a monopoly and few of your customers can escape your fiefdom.  But cloud service providers are setting up their systems at your gates, calling to your citizens with cheap jeans and Coca Cola, and laughing at your titanium shovels.  There is only so long that you can keep your internal customers locked up behind a Berlin wall before the economic gravity of the free market pulls your monopoly apart.

8 Responses

Keep it coming, Scott. Some great thoughts in here. And a call to action to a lot of smart people out there.

  • […] Check out the following article “Economic Theory and IT” at “Pivot Point”. […]

  • Nice analogy. Planning and production of unnecesssary amount of titanium shovels or even sneakers contributed to the collapse of that economy. If you ask a customer how much CPU he/she is going to need, of course they will say some safe ghz. It’s not about giving titanium vms, it’s about giving titanium quality service to customers, and how you do it.

  • Well said Scott. One of the best reads my eyes have read in a long time.

  • Love it. I like to say half-jokingly, half-seriously that IT service management is often indistinguishable from Soviet style planning. In IT, there’s a tendency to think in terms of ‘what the business needs’ in the complete absence of any real feedback loop or mechanism to check whether what you’re supplying is what the business is really demanding or vice versa.

    The standardization mandated by many ‘maturity models’ can serve to make this worse in a lot of instances. IT decides what it thinks is best for the business and then works to supply it – whether that’s what the business wants or not. Add to that the experience of virtual queues in the rain for a loaf of bread (three months to provision a server) and you have something that sounds a lot like the end-days of the USSR.

    The last time this happened, the business went around IT’s back and introduced open systems. We’re seeing a similar situation today and unless IT starts to rely on the only reliable feedback mechanism we know about (prices) to match supply and demand, the problems we’re seeing will only get worse.

    On the bright side, it’s massively simpler to introduce effective pricing than it used to be. Here’s working for the victory of the invisible hand over the five year plans in IT!

  • Fantastic perspective….and even better as it merges two of my favorite areas (economics and IT). I remember reading fat history books about the KGB when I stayed with my uncle as a kid.

    Now here’s a question….how do you value the intangibles? Can we measure everything in IT? (the reliability of internal resources we trust, the value of external resources we trust, etc.) One theme in that to me is trust — cost is one measure of value….and while you may have an SLA attached to that cost (where an SLA is almost a measure of trust), SLA’s are only worth so much (the “virtual paper they’re not printed on” maybe) when you’re in the midst of downtime. I think there will always be a premium for trust while for enterprises will come somewhat second to the cost metric (take a look at the cost per GB of enterprise storage as Exhibit 1).

    And that reminds me….I need to read the Wealth of Nations again sometime soon….

    • I have asked nearly every customer I have met in the past three months if they are implementing Chargeback. Few are and most cite the complexity of the process (not the tools) as blocking their efforts. Amortizing hardware costs over a varying number of VMs seems to be the toughest issue.

      I have sometimes suggested that VM cost might decrease when some features are enabled (like FT or SRM, as an example) since they decrease the energy required to manage hardware failures. This is somewhat like the trust you describe. But when “trust” decreases the charged cost, but its implementation results in a higher VMware license, I see customers’ foreheads wrinkling.

      Sticky space…