|
|
|
KEYNOTE ABSTRACTS
Keynote 1: Scalable Management for Global Services
Ken Birman
Cornell University, US
As enterprises go global, it's increasingly clear that we need a standard Internet service to play such roles as tracking the nodes on which applications are running, monitoring health (live, failed, or unreachable), supporting locking and other forms of synchronization, maintaining parameters and dynamic status, etc. Such functionality could open the door to a wide range of autonomic behaviors. Lacking it, developers have generally rolled their own solutions. For example, the DNS can perform some of these functions, but it lacks the performance, scalability and consistency properties needed for many purposes. This talk will touch on some recent success stories in the area, then look to the future and ask whether it isn't time to replace the DNS with some form of globally deployed management service: a GMS.
Short bio:
Ken Birman is Professor of Computer Science at Cornell University. He
currently heads the QuickSilver project, which is developing a scalable and
robust distributed computing platform. Previously he worked on
fault-tolerance, security, and reliable multicast. In 1987 he founded a
company, Isis Distributed Systems, which developed robust software solutions
for stock exchanges, air traffic control, and factory automation. The
author of several books and more than 200 journal and conference papers,
Dr. Birman was Editor in Chief of ACM Transactions on Computer Systems from
1993-1998 and is a Fellow of the ACM.
|
|
Keynote 2: Data Aggregation over Networks via Integration
Robert Ghrist
University of Illinois at Urbana Champaign, US
This talk considers the problem of aggregating coarse redundant data over networks, using a non-standard integration theory. A toy problem to be addressed is as follows: given a collection of targets in a domain and a network of sensors which count the number of targets nearby, the goal is to determine the total number of targets. The difficulty is that the sensors are minimal --- they can neither
localize nor identify the targets. This makes it difficult to eliminate redundant data.
The solution we construct is an integration theory based on sheaves and the Euler characteristic (both of which will be exposited in the talk). It is powerful enough to solve a wide variety of problems while remaining computable quickly in a distributed manner.
Keynote 3: The Promise of Self-Adapting Equilibrium
Mark Burgess
University College Oslo, NO
How should we understand autonomics? As software engineering, as
biology? Mark Burgess is the visionary author of cfengine, probably the
first autonomic system for server management dating back to 1993, which
now manages hundreds of thousands of computers all over the world. Since
writing his manifesto "Computer Immunology" for self-repairing computing
in 1998 he has led research efforts at Oslo University College to
realize self-healing systems in practice, using strong scientific and
engineering principles. In this talk he will share with us some of the
the principles that have made cfengine successful and his vision for
realizing autonomic computing, including the importance of promise
theory and dynamic equilibria that not only offer engineering
principles, but also reveal the essential economics behind cooperative
computing.
|
|
|
