The DISCO Lab is one of the BRC Research Laboratories.
We are interested in distributed computing systems, distributed applications
(real-time signal processing for example) and performance evaluation of
distributed systems.
Description
The research group works cooperatively on projects involving a variety
of topics in distributed and parallel computing. These include:
Performance Analysis Techniques/Methodologies for Parallel and Distribute
Computing Systems: constructing engineering models (e.g. performance models),
and using these models to predict and optimize performance.
Software Performance Engineering: the development of software design
methodologies which incorporate the use of performance modeling to satisfy
performance specifications (as in a real-time system), or to optimize performance
(as in a parallel computing application).
Real-Time Distributed Computing Systems: cooperative work with industry
in different applications such as high speed non-destructive testing of
railroad tracks and pipelines
Analytic Modeling of Computer and Distributed Queueing developing and
enhancing matrix techniques for understanding the unusual behavior of certain
queueing systems. This has come to be known as "Linear Algebraic Queueing
Theory" (LAQT). Applying these techniques to telecommunications networks,
where many researchers have found that the traffic seen on both inter-
and intra-nets, is very unusual. This behavior, variously known as "Self-Similar,"
"Chaotic," and "Long-range," is not well represented by the standard models
that have been in use for the last 30 years. Our model (called the N-Burst
process) promises to be very useful for understanding why telecommunications
appear to be so unstable. It is the only analytic model (as opposed to
simulation model) that treats traffic flow as the point process it is (as
opposed to `continuous flow' models).
The DISCO Lab is located in the BRC-38 on the
UConn main campus (Phone: 486-0849).
Follow us. You will know the answer to.....
"About the
Distributed Computing : What is it and how can you evaluate its performance?"
