Recent technology advances have enabled wide area application (WAA)
processing with WebSources that are accessible via the WWW. One
challenge to query processing in wide area environments is the
unpredictable behavior of WebSources, depending on the network and
server workloads. The workloads are often affected by parameters such
as the Time of Day, Day, etc. Another challenge is that autonomous
WebSources may not provide metrics needed for accurate cost
estimation. To develop an Access Cost Catalog that addresses these challenges,
one must rely on query feedback to gather cost information and
statistics. The Catalog should continually
monitor the performance of WebSources, and learn from query
feedback to improve the accuracy of its predictions.
We describe a case study in the development of an Access Cost Catalog for
WebSources. We document our experiences in validating this Catalog
and note successes and lessons learned. The Catalog uses
the WebPT - Web Prediction Tool - for learning and prediction.
Finally, we develop
optimizer strategies to meet performance targets (PT) in noisy environments.
A PT sensitive optimizer may be optimistic and ignore expected delay
or be conservative and respect delay. We use a utility function to
measure how well queries meet a performance target.
for recent unpublished papers.