Abstract: A method of enabling transaction probes used to monitor a transaction or modify a primary application handling the transaction. The method begins with retrieving a token associated with the transaction. The token contains information regarding which transaction probes from a plurality of transaction probes will be enabled with respect to the transaction. The token is then read to determine the set of transaction probes from the plurality of transaction probes that will be enabled. The determined set of transaction probes is then enabled.

Abstract: A first failure data capture mechanism receives threshold violation events. The end user may configure which threshold violations would trigger first failure data capture. A correlator may be used to select only the related log and trade data to fit the specific application. When a predetermined threshold violation event is received, the first failure data capture mechanism gathers the appropriate log information. The first failure data capture mechanism may also query for other information related to the transaction that caused the threshold violation.

Abstract: A method, system, and computer program instructions for using existing performance monitoring solutions to detect performance issues in an enterprise, and providing and executing a corrective action on any server being monitored in the enterprise to correct the performance issue. When a management agent on a monitored server detects a threshold violation, the management agent sends a violation event to the management server. Upon receiving the violation event, the management server distributes a corrective action associated with the threshold violation to a set of defined management agents involved in the transaction. Each management agent then runs the corrective action to remedy the performance problem.

Abstract: A plug-in program for monitoring an application having a set of threads. A parameter is associated with each thread. The plug-in program causes an individual thread associated with the application to be terminated if a corresponding parameter violates a threshold. The thread is terminated without interfering with the execution of the other threads within the set of threads.

Abstract: A thread monitor identifies and interrupts locked threads. A byte code insertion tool inserts timers around classes that are bound to threads. Timers may also be inserted around particular portions of code. When a portion of code, such as a thread, a loop, or a condition block, for example, to be monitored begins, a call is made to a thread monitor to start a timer. If the portion of code ends normally, the timer is removed. If the timer expires, then the thread monitor may interrupt the thread or send a notification to a user so that appropriate action may be taken.

Abstract: A mechanism is provided for just-in-time instrumentation that uses bytecode insertion to modify a class definition to store a reference to the static class information to be used by probes. A hook is inserted to provide a callback to one or more probes. When the bytecode inserted hook is executed, the static class information is used to locate and execute the application probes.

Abstract: A system and method to allow an individual transaction in an instrumented software application to be monitored according to its own configuration. In one preferred embodiment, each transaction is associated with a token. The token includes the monitoring configuration that should be used for the associated transaction. When the application entry point is reached for the transaction, the token is looked up and retrieved, and the monitoring policy configuration of the token determines what data is gathered about the transaction.

Abstract: A monitoring program for online transactions can define an ‘enable module’ that requires minimal overhead. When the monitoring program is triggered, the enable module is loaded without a context to determine if the probe is enabled, with the loading of any context deferred until enablement is determined. When the full probe is not needed, the associated overhead of loading a context is avoided.

Abstract: The present invention addresses the problem of linking cross-process and cross-thread subtransactions into a single user transaction. The mechanism of the present invention employs bytecode inserted probes to dynamically detect out of process correlating tokens in an inbound request. The bytecode inserted probes retrieve the correlating token in the inbound request. Based on the correlating token retrieved, the bytecode inserted probes are then used to dynamically determine if the inbound user request should be recorded and linked to a transaction that began in another thread or process.

Abstract: Mechanisms for graph manipulation of transactional performance data are provided in order to identify and emphasize root causes of electronic business system transaction processing performance problems. A system transaction monitoring system, such as IBM Tivoli Monitoring for Transaction Performance™ (ITMTP) system, is utilized to obtain transaction performance data for a system. This transaction performance data is stored in a database and is utilized to present a graph of a given transaction or transactions. Having generated a graph of the transaction, and having identified problem conditions in the processing of the transaction(s), the present invention provides mechanisms for performing graph manipulation operations to best depict the root cause of the problems.