Databases in Grid Applications: Locality and Distribution

Speaker  Paul Watson

Affiliation  Newcastle University & North East Regional e-Science Centre

Host  Dan Fay

Duration  01:01:39

Date recorded  29 September 2005

This talk focuses on two areas that experience in building database-oriented e-science applications has shown to be important. Firstly, methods of promoting data locality are vital due to the high cost of moving data in service based distributed systems. Databases provide an excellent basis for achieving this due to their potential for moving computation to data. I will describe a new infrastructure that further promotes locality by enabling service-based computations to migrate to data. Secondly, the ability to combine information from a set of distributed databases has proved invaluable in many applications. I will describe the design of an adaptive distributed query processing system that is able to exploit the facilities offered by an underlying grid infrastructure. This includes dynamic re-optimization as the pool of available resources changes, and the ability to tolerate node failure.

The talk will also include a brief overview of research at the UK's North East Regional e-Science Centre, which Paul Watson directs. This includes work on bioinformatics, distributed databases, scalability and virtual organizations.

©2005 Microsoft Corporation. All rights reserved.
> Databases in Grid Applications: Locality and Distribution