Tracking Transaction Footprints for Non-intrusive End-to-End Monitoring

Existing transaction monitoring solutions are either platform-specific or rely on instrumentation techniques, which limit their applicability. Consequently, transaction monitoring in enterprise environments often involves the manual collation of information spread across a variety of infrastructure elements and applications, and is a time- consuming and labor-intensive task. To facilitate self- governance in enterprise environments, we present an online, non-intrusive and platform-agnostic solution for transaction monitoring, where the only inputs required are (a) system log files in which footprints left by ongoing transaction instances are recorded, and (b) a model of the transaction, in terms of the valid sequences of steps that a transaction instance may execute and the expected footprint patterns at each step. Given these, our solution generates a dynamic execution profile of ongoing transaction instances that allows their status to be tracked at individual and aggregate levels, even when transaction footprints do not necessarily carry correlating identifiers as injected through instrumentation. We describe our monitoring architecture and algorithms, results from an empirical study, ongoing work on run-time transaction model validation and directions for future research.