I'm a big fan of so-called serverless architectures. The idea of these products is that developers don't have to think about spinning up servers to do some processing—rather a construct that goes something along the lines of "when trigger A happens, set off process B, and when process B is complete, your job is done" can be enabled.
Amazon Web Services (AWS) was the first of the public cloud vendors to launch a serverless offering, AWS Lambda. Since then, it is an approach other players have followed.
But while serverless offerings add massive value in terms of simplicity and economics, they provide challenges. The servers that run the actual code to process these events are not exposed to developers. As such, developers have zero visibility into how those servers are working and what they're up to.
+ Also on Network World: Serverless computing: Mapping out the possibilities +
Traditional approaches towards server monitoring—by way of agents and the like—cannot apply to serverless architectures. Lambda functions can be triggered via an application program interface (API), through in-app actions or by AWS events, such as a new object being added to an Amazon Simple Storage Service (Amazon S3) bucket. Since launching, Lambda adoption has grown quickly, with large enterprise companies such as Netflix leading the way.
So, it is interesting to see an announcement from cloud monitoring vendor Datadog today that they are rolling out an integration with AWS Lambda. What this means is developers can emit custom metrics specific to their Lambda functions, right from the functions, on top of monitoring the metrics provided via Amazon CloudWatch. Once collected, these metrics can be used to create visualizations and alerts alongside operational data from the rest of the infrastructure environment for actionable insight.
“Serverless architecture is ushering in a paradigm shift in cloud computing. Developers no longer have to provision or manage servers to execute their applications, and companies only pay when code is executed,” said Daniel Langer, product manager at Datadog. “With Datadog’s AWS Lambda integration, users can gain insight into their serverless application’s performance that was previously inaccessible.”
My POV
This is an interesting piece of news for a couple of reasons. First, it is an indication of just how seriously customers regard these serverless approaches. It is only a couple of years since Lambda was introduced, but it has quickly moved on from being a very cool, but difficult to understand science project and is now fairly widely understood and adopted.
The second indicator is one relating to Datadog: This is still a very young company but one that has absolutely grown like the weeds. From very modest beginnings only a few short years ago, Datadog has grown to be seen as one of the more established cloud monitoring platforms. Indeed the list of AWS products it integrates with pretty much spans the entire AWS continuum.
Of course, that is necessary given that AWS offers its own monitoring solution, CloudWatch. But it is the added value on top of the basic metrics that Datadog is looking for, and it will be interesting to see what they do with the addition of Lambda information.