|
|
Thesis: "Capacity Planning
for Database Management Systems Using Analytical Models"
|
|
Abstract (full text
not available)
|
|
In this thesis, we propose
a scheme for performing capacity planning for database management systems.
This scheme uses an analytical model. The model can be incorporated
into a software tool that allows users to test database system tuning decisions
for DB2. The user would pick resources from a library, and specify
a hardware configuration on which DB2 would run. This tool returns an
estimate of the performance indices of DB2 on such a system. This software
would save the user from actually having to try out design decisions, which
can be costly and impractical. The software can be used to find the
performance bottlenecks. The software can also be useful for a user
whose system is currently performing within the required service levels but
the workload of the system is expected to increase and could cause a degradation
of the service levels. The first step in our study is to monitor the
performance of the system on which the database management system (DB2)
is running. We use NT and DB2 performance monitors to measure the
utilizations of the system resources. Then, we convert the monitored
values into parameters for a queuing network model that represents
the DBMS (DB2). Using curve-fitting techniques, we formulate these
parameters. And finally, we use the MVA technique to "solve" the model
and to calculate the performance indices.
|
|