How Buffer Pool Analyzer is used
The suite of Buffer Pool Analyzer tools can be used to monitor, analyze, and tune buffer pools.
- If frequent monitoring and performance observation with minimum effort is the goal, high level summary reports can be created on the host from collected performance data. This process can be automated by frequently running batch jobs. Data collection and report creation can be configured for individual needs. Optionally, data can be viewed on the Buffer Pool Analyzer client in attractive and intuitive graphical representations.
- If an analysis of buffer pool related problems is needed, several summary and detail reports can be created to quickly identify possible problem areas. Reports can be customized to provide timely and content-specific information.
- If optimization and tuning of buffer pool resources and
usage is the goal, the object placement and buffer pool sizing tool
and the buffer pool simulation tool are first choice. Based on real
and representative buffer pool performance data, these tools ease
the process of finding optimal use of buffer pool resources and simulating
the effects of possible changes.
- The object placement and initial buffer pool sizing tool uses predefined and modifiable expert rules and the objects actual access behavior to calculate optimized buffer pool arrangements. It recommends ready-to-use SQL ALTER statements and DB2® ALTER commands, with their parameters set to the recommended values. The tool can be used to balance buffer pool sizes, for example to separate sequentially from randomly accessed objects into different buffer pools, to optimize memory usage, and to improve application response times.
- The simulation tool uses actual objects' access behavior and simulates different object placements and buffer pool size ranges. The simulation results provide a reliable prediction about the effects that different placements and sizes would have on a system. Simulation is used to perform what-if scenarios to balance buffer pool sizes and performance and to provide precise information about the prospective effects of different buffer pool scenarios.
Both tools complement each other by performing the (often complex and iterative) task of optimization and tuning on a client, thereby still relying on actual performance data. The strength of object placement and initial buffer pool sizing is its rule-based algorithm and its ready-to-use recommendations. Simulation takes the surprises out of planned changes and minimizes the number of system disruptions.
The long-term analysis function adds another dimension to monitoring, analysis, and tuning: historical and current performance data can be combined and analysed as a whole to easily detect trends, hourly, daily, and weekly peaks, repetitive performance pattern, unbalanced resource usage, and much more. The client-based long-term analysis function provides an array of intuitive selections to focus on important performance indicators, buffer pools, and database objects.