Datadog puts Profiling on a long leash for initial trials

Datadog puts Profiling on a long leash for initial trials

Monitoring service company Datadog has started a private beta for Java, Python, and Go developers with an interest in adding profiling to their infrastructure.

Profiling, as the new service is called, does what it says on the tin, giving developers a way of analysing stack traces to investigate performance issues and find bottlenecks in an application’s code. Beyond mere error hunting, this can also be helpful to reduce latency and infrastructure cost.

The tool is integrated with the rest of the Datadog stack, which supposedly gives users a way of easily switching between application monitoring and profiling for example. Amongst other things this is meant to support them in getting to the cause of particularly slow requests quicker.

Devs that look into profiling as a way of optimising their application get pointers on that in an analysis view. It lists major inefficiencies found during a code analysis run. However, users can also dive into key metrics such as CPU usage, memory allocation, and a variety of I/O operations on their own.

Depending on the problem at hand and the language used in the application, there are also ways to filter by attributes such as threads or packages. Additional features such as long time performance observation allow ways of figuring out if issues follow a temporal pattern or are just outliers.

Since it’s still in closed beta, access to Profiling has to be requested via a special form. Right now the offering only works for Java, Python, and Go code, but Datadog promises to add .NET, Node.js, PHP, and Ruby to the list soon.