Virtualization Automation – A Proper Virtual World

Know What You Don’t Know?

August 1st, 2007

Virtualization? What The Heck Is That? – Chapter 02

Know what you don’t know. Any experienced computer technology expert knows this simple rule, if they don’t fire them now, seriously. While you are at it, if the company CPAs do not know this as well, fire them too, I will explain later why. This know-what-you-don’t-know concept is the basic principle of the scientific method. In other words, if you don’t know what questions to ask in solving a problem, you have no possible hope to finding a viable solution. How does this prove applicable to virtualization?

Decentralization of resources, in a typical multiple server meshed network, which is the modern corporate infrastructure model that dominates today, is driven by project requirements. Infrastructure grows project by project. Thus the total infrastructure is hard to manage and track. Asset control, resource allocation, end of life budgets, infrastructure costing, etc., are project focused for approval, and costing at a site level or greater is almost an after thought. Short of being a CPA at the corporate financial office, almost no one sees the 10,000 foot view, with few ever seeing the 100,000 foot view. But virtualization requires the 100,000, the 10,000 and even the 1,000 foot view concurrently and predicatively! Thus, you need to know what you do not know, past, present, and to a reasonable degree the future of your computing infrastructure, to efficiently and effectively implement virtualization from a scale and cost perspective.

For Power, cooling, connectivity, server hardware, site infrastructure, personnel, etc., what are your fixed costs? What are your variable costs? Just looking at hardware, for example, how many servers did you purchase 3 years ago? 2 years ago? Now? Next year? Next 3 years? Next 5 years? Don’t laugh! In a project driven justification model for purchase approval, these predictive scenarios are not straight forward, and drive CPAs insane. Forcing financial analysis in a future context is an educated guess, which CPAs hate, even if they don’t say so. Telling management you have to spend $15 bucks to save $10? And the very first thing the CPAs ask is how do you know only 15$ is needed? Then they ask, how do we charge back the $20 (they never believe it is only 15) across 40 entities, which start at 10 and grow into 40, since you have 40 virtual machines online today? By CPU utilization? By DASD utilization? What statistic? CPAs, thus management, never focus on the $10 bucks in savings. This exact issue has happened to me enough times, over the last few years; I take aspirin to the conference room, and leave it out on the table for any one that needs it. I think Las Vegas has standing odds at 3-to-1 any CPA in the room goes for the aspirin first. I figure you fire CPAs until they suggest up front a predictive graduated cost model for pre-provisioned virtualization funding, soon or later the word will get out new thinking is required for virtualization costing.

If you don’t know what you have done in the past, and can’t understand what you really need in the future? Then deciding to implement virtualization is counter intuitive, and impossible to justify.  Remember the comment about firing the CPAs? What company, large or small, really knows what it is spending on infrastructure? What it is actually being used effectively in the infrastructure? Most CIOs in fortune 100 firms know the budget numbers, but not the lines items, and almost no one knows the efficiency-of-use for said line items. It can truly be said, for virtualization, the devil is in the details, past, present, and future.

Entry Filed under: A Proper Virtual World, ToutVirtual, Virtual System Management