Datadog released a new machine learning monitoring capability, called Watchdog, and added trace search and analytics capabilities to its network monitoring platform. Watchdog leverages machine learning (ML), algorithmic learning, and dashboards to automatically identify issues in the network.
Datadog’s self-named monitoring platform aggregates logs, metrics, and tracing from the entire network stack in one platform. It has an library of over 250 technologies that it integrates with and monitors right outside of the box. The platform includes capabilities for application performance monitoring (APM), anomaly detection, and outlier detection.
According to Ilan Rabinovitch, Datadog vice president of product management, Datadog wasn’t as intuitive as it could be. “With the things that we offered in the past, we could apply any of those functions — anomaly detection, outlier detection, forecasting, any metric in our platform, even your APM data — and you could show that on dashboards or create an alert, but you had to know what you wanted to look at,” he said. “You have to go out of your way to set that up.”
Using algorithmic learning and ML, Watchdog tries to improve this process by automatically looking at all the data sent by an enterprises’ infrastructure and applications. The machine learning algorithms identify interesting, relevant patterns to an enterprise’s specific environment.
Rabinovitch said that the differentiating factor of this capability is that “it just works right out of the box and it starts sending data. We’re immediately looking for those patterns to let you know what’s interesting and relevant to you.”
The algorithms are developed by Datadog’s team of data scientists and engineers using application and service modeling. They use data from the algorithms behind the platform’s existing anomaly, outlier detection capabilities and forecasting functions. The exact math though? “At the moment that’s a bit of our secret sauce,” said Rabinovitch.
Datadog also announced the availability of a Trace Search and Analytics tool that searches through an enterprise’s volumes of application and performance data. While it ties into the entire monitoring platform, it is specifically tied to its APM and Distributed Trace tool.
According to Rabinovitch, this is like a search-engine tool that enables enterprises to explore trace data on-the-fly. “You’re able to see all that data together,” he said, adding that enterprises can get all the context from all the pieces of the network in on place, without having to “bounce around, even between parts of Datadog, or between Datadog and some other product that you may have.”