SOAs for Emergencies

I just got a note from David Stephenson of Stephenson Strategies alerting me to his blog entry which points to an issue of my Network World Web Applications Newsletter titled Enterprise customers make good use of wikis to combat e-mail overload (whew). David’s piece discusses how to build and emergency communications system and the reference to the Web Applications Newsletter is due to the discussion of an SOA called SocialText that several enterprises have used to mitigate the flow of e-mail. The whole topic of systems to support emergency operations is crucial in IT but of course it is one thing to be thinking in terms of your company’s headquarters getting flooded and quite another to be dealing with the consequences of a hurricane hitting a city, county, or state as it happens …

For the kind of tasks that David is concerned with the whole concept of usability is very different from normal IT requirements. The biggest danger with general systems used to support groups managing time-critical functions is the potential for information overload – as stressed users dump their observations, requests, recommendation, and thoughts the pool for live data will grow rapidly. Combine that flood of data with news feeds and raw data (weather reports and so on) and the need for an information management hierarchy seems obvious. What these systems need are improved data classification engines to monitor what a user puts into the system and then brief him on “stuff” like the stuff he deals with and make sure his stuff is routed to the people who it matters to. What these kinds of systems need is to go further in understanding the user’s relationships with other users such as who he reports to, who reports to him, who he works with, who else he communicates with, and so on. It is not enough to specify the hierarchy there’s also the need for systems to “understand” when the hierarchy isn’t being used or wher it is being used in unexpected ways. For example, during the Katrina disaster there was a short circuiting of the chain of command that should have been known about. Michael Brown, then head of FEMA ignored his boss, Secretary Chertoff, and went around him under the belief that he could get things done more efficiently that way. Whether you believe Brown did the right thing or not the fact is that it should not be possible for such a situation to occur – the command and control system should make communications transparent and auditable. When you start to think of what it takes to deal with disasters on a national scale it makes dealing with a fire in your data center look pretty easy, doesn’t it?