One of the fastest growing use cases for machine learning is providing insights into how software releases in a continuous delivery cycle affect overall IT operations. Keeping with that theme, software-as-a-service (SaaS) vendor Perspica Inc. has just released a new analytics platform for DevOps aimed at providing greater insights into the operational impact of continuous software delivery.

San Jose, Calif.-based Perspica said last week its integrating DevOps stacks into its operations analytics engine. The enhanced platform provides predictive analytics and root cause capabilities for application infrastructure stacks, diagnostic, monitoring and troubleshooting tools.

Perspica has also added support for new DevOps platforms including Amazon Web Services, Collectd, Docker containers, Flume, Kafka, Loggly, MongoDB, OpenTSBD, Redis and Zookeeper,

The company positions its Incident Replay platform as a “time machine” that’s able to track performance data, logs and topology. Using the platform, DevOps teams can investigate the impact of a new code release on application performance, helping them to pinpoint any problems that negatively impact performance.

“The continuous delivery model means developers have to move quickly, and they are hampered because they often can’t see how the new software release is affecting operations,” said Perspica CEO Dan Maloney in a statement. “This can lead to application performance issues and even outages.”

With the replay capability, DevOps teams gain greater visibility into the impact new software releases have across “the entire development stack”, Perspica claimed.

The company further touts its platform’s ability to help developers visualize the state of their IT infrastructure based on performance data, logs and topology changes. It’s possible to compare current operations with past performance too, delivering an expanded ability to detect problems with new code and limit the impact of outages.

Perspica’s Big Data-centric platform is able to ingest the millions of “events” and performance metrics generated by hyper-scale infrastructure each second in real-time. The platform also leverages machine-learning techniques to get a handle on the network topology by determining the relationships between different components. This automated approach helps Perspica to determine what it should consider as “normal behavior”, and thus alert developers to anything unusual that’s happening.

Any anomalies will be detected instantly due to sudden changes in the way network objects behave, at which time the platform automatically kicks off a root cause analysis to determine what’s gone wrong. It then follows up with “intelligent recommendations” to handle the problem.

Perspica’s platform is also armed with a predictive analytics tool that can help to spot early warning signs that may lead to an outage, and recommend actions to prevent it.

Photo Credit: horandarwin85 via Compfight cc