Here are some case studies from the Jerrold Patz files that
have been selected for discussion. Each reveals a different facet of Behavior Modeling.
These examples cover SUN Unix and HP Unix capacity planning and tuning, and LAN
modeling and tuning. These cases were selected to reveal the scope and depth of
behavior modeling projects.
The graphs may appear to be deceptively simple: They are
designed that way. In fact, each graph
is the result of data acquisition across many hours/days for many variables. The
data are then modeled and the
graph is a physical representation of those models. The data representation
allows for an understanding
of the underlying issue for each Case Study.
Click on any Case Study or the Menu Bar above for additional
Case Study #1
This project involved a large, underperforming SUN configuration.
When the customer sought a complete system upgrade, we recommended that the
system be studied first to determine what needed upgrading. Rather than a
forklift upgrade, the customer achieved desirable performance through system
Bottom Line: The customer saved over $300,000 by not
buying unnecessary upgrades.
Case Study #2
This project involved a LAN that was experiencing poor
performance. We recommended that the system be studied to determine how the LAN
was functioning and what, if anything, it need to perform better.
Bottom Line: Rather than upgrading hardware, tuning
delivered satisfactory results and increased user satisfaction.
Case Study #3
This project involved an underperforming HP
configuration. When a consultant recommended a complete system upgrade, we
were engaged either to validate the recommendation of the first consultant,
or to suggest a different course of action. Again, we recommended that the
system be studied to determine what needed upgrading.
Bottom Line: Customer saved over $250,000 on
an unnecessary system replacement and still achieved desired performance.
Case Study #4
This project involved an underperforming WAN. We constructed a network
behavioral model and determined that there were several sub-optimized
Bottom Line: Substantial performance gains on
the WAN, increased network satisfaction and
reliability. All of this was accomplished through tuning with no hardware