Information Technology Reference
In-Depth Information
F
IG
. 12. A sample SRGM fitted to KDE data.
6 . 5
G e n e r a l O b s e r v a t i o n s a b o u t W e b R e l i a b i l i t y a n d Te s t
E f f e c t i v e n e s s
As demonstrated in this section through case studies, web software reliability and
test effectiveness can be evaluated based on information extracted from existing web
server logs. Our key findings and related observations are summarized below:
Measure derivation and data extraction
: Four workload measures, bytes trans-
ferred, hit count, number of users, and number of sessions, were derived in
this section for web workload characterization and reliability evaluation. Hit
count, byte count, and user count can be easily extracted from access logs, due
to their direct correspondence to access log entries and the data fields “bytes
transferred” and “IP address.” Session count computation may involve history
information for individual users or unique IP addresses, but properly identified
user sessions with appropriate time-out values can reflect web usage better than
simply counting the users. Detailed failure data can be extracted from error
