Policy-driven computing relies on user behavior and a host of preceding technologies
Aligning technology and business — what a concept. Everyone is talking about it, but to put things in perspective, I imagine the day after the wheel was invented the wife of the inventor got sick and tired of watching her husband roll it down the hill all day. She probably said, “Cyxny, honey, why don’t you make another one and put them on the cart?”
David Luckham, professor emeritus of electrical engineering at Stanford University had a similar eureka moment back in the mid-’90s. Luckham and his research team at Stanford were building simulation systems. The idea was to build a model of software architectures and execute on test data that produces lots of events to analyze behavior.
Luckham realized it might be commercially viable to decouple the analysis tools from the simulator and apply the analysis to any system that creates events. Such was the birth of CEP (complex event processing).
In a parallel universe, network management systems were the precursor to BAM (business activity monitoring). The need to keep tabs on activity within a given enterprise kept rising from the application layer to the business process layer and inevitably to the management layer.
Now, the merging of CEP and BAM brings us to something called policy-driven computing. Companies are trying to codify their business policies in computer systems so that these polices can be considered when certain processes are executed, says Evangelos Simoudis, a partner at venture capital company Apax Partners.
For instance, take a balance transfer from my old credit card company to the new company that spent a lot of money convincing me to switch. The new company wants my money as quickly as possible and needs to monitor the set of complex events surrounding the transfer of funds. So it must create business rules that also take into account the temporal component within the rule to be sure everything is happening in a timely manner.
I spoke with Keith Blackwell, chairman and CEO of Bristol Technology and one of the pioneers who will use BAM to take policy-based computing to the next level. He explains that Bristol’s Transaction Vision uses nonintrusive software components to track the behavior of transactions across applications. As the transaction splits into various pieces it builds a tree of events to aggregate the events back into a single transaction at the end of the process. The system tracks both the IT operations and the business content.
The second component of Transaction Vision uses an analyzer called Amit, licensed from IBM, to collect all of the data and apply policies to the transaction. It looks for patterns or exceptions and sends an alert to designated business people when there is a problem.
This is indeed the alignment of business and technology. But keep in mind that rules that enforce policies are based on recognizing patterns of behavior. Computers don’t do that. Yet.
If you’ve stopped for gas recently you may have noticed that when using a credit card at the pump it usually requires you to input your ZIP code. This is the result of someone at a credit card company noticing a pattern: Thieves who steal credit cards often use them at gas stations first.
“In the final analysis you have to learn by watching,” Luckham says.
It’s good to know some things never change.
