A Content Oriented Analysis Approach
Reading java system dump is a daunting task, Cognos Extension makes it easy. The following is an introduction, I will follow up with some concrete examples.
IBM Business Analytics Extensions is a set of queries for IBM Memory Analyzer, to analyze the following issues,
- Application server memory usage planning and rationalization
- System stability related such as slow performance, performance degradation and hung issues
- Resource utilization sampling, system tuning and optimization
- Configuration validation and adjustment
- Non reproducible exceptions
- System crashes
- General Java system health check
Content Oriented System Dump Analysis
Actively promoted and advocated by the engineering, quality and delivery (formerly L3) group, the content oriented system dump analysis focuses on system content, or and data structures analysis rather than software source code. The approach systematically reveals the complex data structures and system process in a way similar to the operational view of a production system. Without the need to know a product’s source code, administrators and support analysts are able to investigate issues by focusing on system and user behaviors. This allows support staff to work around issues with little support from software developers. Ultimately, this will let clients pass the identified use case to the development team for an official fix. The approach greatly reduces dependencies on the development team and speeds up the problem resolution cycle.
The content oriented approach focuses on the operational view, revealing the following information:
- System topology, including what services are available on what server instances in a distributed environment
- System settings, including system configuration, system parameters and advanced settings
- Client requests consisting of user actions. For example, a http request, a soap request, a jmx request, an axis request etc.
- Various elapsed times of a process, i.e., the start of a request or intermediate elapsed times during a request process life cycle
- User activities, a summary of all on-going requests in the system, with high level abstraction of a user action, such as who a user is, from what browser with what ip address, elapsed time, and what request is doing, etc.
- Key connection and queue size information in between components
- Utility tools to search and display well known objects, for example, a request and its elapsed time
You can download MAT via the following link. Cognos Extension is bundled there, however a bit out dated. Open a customer case to request the latest.