Abstract: A constant monitoring and recovery system that enables the measurement of task usage/metrics and performance of software-based business applications. These metrics can then be evaluated by logic, in combination with each other and/or a timer, in a distributed environment. This is accomplished with the introduction of an extremely low overhead to the application host. The results of the logic evaluation can cause real-time and real-world responses such as application restart, interaction with load balancing equipment, or notification (via email, pager, etc.). Data can also be persisted to a database for later archival or evaluation, with the ability to reduce granularity over time. Specifically, an application manager monitors business application processes, notifies failure and automatically recovers software based business applications based on the logic of the underlying applications.

Abstract: A target software behavior pattern recognition and analysis tool is disclosed that operates during the real-time execution of monitored target software on a computer, derives a subset of measurements indicating the target software performance and activity, automatically derives a baseline from the subset of measurements, and establishes a state of health of the target software based on comparing real-time measurements to the baseline. The monitored target software contains an agent for deriving a subset of measurements indicating the target software performance and activity. The agent includes sensors embedded into the target software to measure specific code segments, and an examiner which receives measurements from the sensors and transmits these examiner values to a central server. The central server automatically derives a baseline from the received examiner values, and automatically establishes a state of health of the target software based on comparing real-time measurements to the baseline.