A San Francisco-based startup is having its coming out party with the broad release of Java application performance management software, which is designed to troubleshoot problem transactions in distributed environments.

AppDynamics version 2.0 became available today. It is the first non-stealth version of the product, and it adds new event-type monitoring and cloud computing provisioning capabilities that were previously unavailable to customers.

AppDynamics was founded in 2008 by alumni from Oracle and Wily Technologies. Its objective is making an APM solution that works across a mix of open-source and proprietary components, not just a pure WebLogic of WebSphere stack, said AppDynamics’ vice president of marketing Steve Roop.

The company focused its technology on monitoring transaction health across distributed systems, and it wanted to make its APM technology agile-friendly, Roop added.

Pricing starts at US$5,000 and is based on the number of JVMs deployed into production.

AppDynamics works by instrumenting entry and exit points into the Java Virtual Machine to analyze execution flows and threads, said AppDynamics founder Jyoti Bansal. “There is less than 2% overhead in production,” he claimed.

An automated mapping function discovers application tiers and back-end services as new code has been introduced, Bansal said. Transaction flows are visualized on an interactive diagram that can drill down into lines of code, making it easier for agile programmers to work with.

The traditional approach toward instrumenting software is like a Rubik’s Cube that few people could figure out how to get right, Roop said. It often required developers to sort through raw data to find the root causes of dropped transactions.

AppDynamics 2.0 adds new event types to transaction flow monitoring, including deadlocks, connection pool health, memory leaks, stalls and thread contention, according to the company.

A cloud orchestration component predicts peaks and troughs in demand by setting baseline application behaviors and performance pattern analysis. It can “burst out” (or scale) applications into the cloud or scale down in response to policies, Roop said.