Facing The Customer: Stories from the front lines

Sometimes Sun gets dinged for being too rigid in our standard configurations. The web store has "small", "medium", "large" and sometimes "extra large" configurations that can be modified to add memory, drives, PCI cards and service options. The online list of parts and configurations is not as exhaustive as some of our competitors, although just about any supported configuration can be had if purchased direct or from a reseller.

I have a customer that was telling me why he loved to buy from Dell. "I just go on to Dell's website," he gushed, "select whatever configuration of a server I want, and it shows up in a day or two. Sometimes I buy them five or ten at a time just to have on hand." This is a pretty big customer. In fact, huge would not be an understatement as they are one of the top law firms in the country and globally for certain fields.

I thought to myself, "I wonder what your CIO would think if I told him you said that."

It seems like most customers I talk to are trying to standardize components in their data centers. Managing a different configuration for every machine got old a long time ago as patch management and configuration management received a much needed dose of rigor. More importantly, however, it raises the question of predictability. How can an IT environment have any level of predictability if  purchases are made on a whim? Assuming that purchases are made to support a business initiative, it should be well known what computing resources would be needed and when.

Really everything that is done by IT departments is to ensure predictability in one form or another. Predictability to handle load is called capacity. Predictability of behavior during a hardware error is fault management or fault tolerance. Predictability of staying in business during a disaster is (no, not disaster recovery) business continuity. Process is the part of the IT world that allows people and product to come together in predictable ways. Clustering is a process by which nodes can cover for each other. Predictive self-healing is a process that Solaris 10 uses to predict hardware failures and maximize availability. Virtualization is a process used to predict utilization.

The point is that if there is no process that ties purchasing choices (or any IT choices for that matter) back to business initiatives than predictability becomes very hard to achieve. When it comes time to report the costs of supporting the business back to the CFO, COO or CEO, the level of predictability is directly proportional to the level of confidence the CIO will have.

If we accept this than it's little wonder why the top concern among CxO level executives for 2007 is "improving and demonstrating business/IT alignment." (Forrester's Business Technographics, July 2006.)

I think that business are wise to understand how requirements should drive purchasing and not throw those decisions, along with caution and predictability, to the wind.