Workshop on New Challenges and Directions for Systems Research
St. Louis, MO
July 31 - August 1, 1997
Focus Group on System Support for Emerging Applications
Participants:
- Gary Nutt (organizer), University of Colorado
- Richard Muntz, UCLA
- Raj Rajkumar, CMU
- John Chapin, MIT
There is a new class of applications emerging in contemporary
systems. It can be characterized as a collection
of independent units (threads, objects, etc.) all providing
services to a single user in a workstation environment. Some
units perform traditional tasks, but others must manage graphics,
images, continuous media, etc. A virtual environment
is an example for this class of software.
This focus group addresses new challenges and directions in system
research driven by the requirements of emerging applications.
A set of strawman issues to be considered is:
- Where are the synergistic parts of traditional system disciplines,
and how should they be combined to support this (and other) classes
of applications?
- What are aspects of a computation model and API for these applications?
- There should be a general paradigm shift in resource allocation
strategies -- away from unilateral decision making to decentralized
policies. What is the range of paradigms that should be considered?
- How should scalability be supported from the server/WAN scenarios through
desktop/LAN to embedded/clustered contexts.
- The OS needs to provide more information regarding things
like per process resource usage; how should the OS call interface
change?
- Resource management to satisfy different QoS metrics (specifically
timeliness, reliability and security requirements)
- Distributed admission control for resource allocation in the presence of
incomplete/inconsistent information
- Should this kind of resource allocation be managed by middleware?
(Should middleware be treated differently than other application
software?)
- Distributed real-time object services
- QoS-aware and bandwidth-aware many-to-many communication services
- Co-existence with current protocols
- A new real-time/multimedia NSFnet
Return to: Table of Contents