Monday, April 30, 2001
11:00 AM
CSB 209
Mani Chandy
California Institute of Technology
Distributed Computing for Crisis Management
Crises, by their very nature, are not planned. When a crisis occurs task forces are created to deal with the crisis. Because crises are not planned, the membership and organization of task forces that deal with crises are, likewise, not planned. The organizations participating in a task force may cooperate in the task force while they compete in other areas, and so organizations collaborating in a task force may not trust each other completely. The research problem is this: What tools help in building information infrastructures for task forces that deal with crises?

Since many organizations participate in task forces, the information infrastructure for a task force is a distributed system. A service offered by an organization in a distributed system is a computing or communication function that can be used by other organizations. Central problems in distributed computing include publishing, discovering and binding to distributed services. Publishing means describing a service so that others can discover it and connect to it. Binding means mechanically connecting to a service without necessarily having human intermediaries establish the connection. I'll discuss these problems, industry standards in these areas, and open problems.

I helped to found a company called iSpheres, partially owned by Caltech, to solve this research problem in a commercial environment. One of the goals of iSpheres is to give the software away free to humanitarian organizations that help people in crisis. The company sells its products to the Fortune 2000 for implementing information infrastructures for the "extended enterprise" or "virtual enterprise." Virtual enterprises and task forces have a lot in common. I'll talk about the fun (lots of it) and difficulties (even more) in starting a company with a different vision.