Optimization Techniques Gartner-style

I'm attending Gartner's Symposium ITxpo this week in San Francisco, which does an excellent job of showing the landscape of IT currently and providing insight into trends. Four major themes dominate: the consumerization of IT, alternative delivery models of IT, green IT and the changing shape of IT. Of course, those themes translate into sessions of a more practical nature, such as the one I attended on Monday, given by Bill Rosser, a VP distinguished analyst, on the topic, "Business Process Improvement: Disruptions and Responses."

Although there's much I could share from the session, I wanted to put up a quick posting on business process optimization. What is this precisely? Rosser says it has three characteristics: It involves "applying process intelligence to understand and adjust processes to be the best possible design"; it works well for "determined processes"; and it's a "must for undetermined processes." Why a must? Because the concept of optimization is more than just some clever algorithm. It includes continuous process discovery and intelligence.

Rosser says optimization is a comparatively new area of development that will be growing.

He provided four optimization techniques to consider:

• Inline simulation: Using actual results of process performance vs. design simulation performance. In other words, how is it working in the real world vs. how we thought it would work in simulation? You make a comparison and get new results.
• Correlation: Compare fluctuations by any given timeframe. Then take advantage of smaller fluctuations in preparing for bigger fluctuations.
• Predictive modeling. This involves doing optimization by analyzing history to prepare for an anticipated event in the future.
• Option algorithms. You generally have to deal with conflicting goals. Often, for any given process, you may want fast performance, low cost, and good quality. As you push on one factor, something else suffers. Conflicting goals takes a look at those factors and examines, as you change performance and design, how does that affect achievement of different goals? It also enables you to smooth out tradeoffs between conflicting goals to get the best balance across desired goals there.