Charlie Bess had an interesting post over on the EDS blog that led me to this article in the November IEEE journal titled: Toward the Realization of Policy-Oriented Enterprise Management (fee charged). The article is by Matthias Kaiser from SAP labs here in my home town. He is doing some research with a group at Stanford working on Policy-Oriented Enterprise Management.
The premise of the article, summarizing dramatically, is that the key to overcoming complexity is letting users adapt business structure to suit their needs so that software changes are more manageable. To do this he proposes enterprise services as the tier that wraps business objects and use lower level services. He also talks about a policy-oriented approach to defining business goals in a declarative way and how this leads to the appropriate business process (contrasting this with trying to design a business process and then impose policy on it). While he talks about enterprise policy research I feel he misses the huge progress made in decision management over the last few years. Many organizations out there are using decision management technology like business rules and analytics to state what their business is, monitor its performance, intervene in conflicts and update rules as and when necessary. They are, to a very large degree, managing their enterprise based on their policies. While the bigger ambitions of the project clearly require research, especially those around generating processes from policies, I think a lot of what they are hoping to enable could be designed today using these kinds of technologies. But these disagreements should not distract from the quality of the article.
I checked out the POEM group at Stanford too (http://logic.stanford.edu/POEM). Their mission is to “Allow enterprises to react immediately to changes in business policies.” This sounds so much like one of the key objectives of EDM that it is almost scary. Their approach is to “Use computational logic and semantics to construct a virtual enterprise physics in which the business policies stated by managers act like “physical laws” to ensure that the right actions always occur, as soon as they are added.” While I wouldn’t say you could use EDM to construct virtual enterprise physics, you can use it to react immediately to changes in business policies and to new data as you collect it.
Comments on this entry are closed.
Thanks for both links to the article by Kaiser and to the POEM stanford research group. The topics they tackle in their research are very similarto mine except that I am concentrating on the application of the use of policies to modeling and enforcing compliance on business processes. To that extent I also believe that using semantics for building policies for the enterprise is the way to go since it ebnnables your policy agents to interact on a higher level of abstraction and thus process many additional non-structural information ion order to take decisions. Especially in the field of enterprise modeling, modeling the assets bnecessary to take decisions is not enough when compliance rules require taking complex meta-information about the enterprise resources on which to enforce compliance. This includes business processes of course. Once you have achieved this it becomes necessary to manage change in policies and in compliance requirements in an automated way, otherwise the complexity that compliance requirements represent is going to get too overwhelming to cope with.