Splunk prototyping DevOps efficiency measurement tool

Exclusive Splunk has developed a prototype product to measure productivity in DevOps teams.

The “Splunk App for Build Analytics” has been deployed in a closed beta and is now running, live, as a bespoke service at several Splunk customer sites.

The tool's existence was revealed to The Register today by Andi Mann, Splunk’s chief technology advocate, who described it as a “proof of concept”, but one that is nevertheless being used by actual customers.

“We have got a big chip manufacturer in the US who is pulling data about how much development work they have planned, what is in projects, what is priority one and two, how efficient are those individuals and teams, when they are going to be able to finish this project they are assigned to and if it is not soon enough how much will it cost to get extra contract resources to make the date that they want.”

Mann said a one-off, custom deployment of the tool can “go beyond people and allocation of resources and look at quality metrics, bringing in data from thinks like SonarQube, Selenium, AppScan, Fortify - the automated testing tools - then starting to correlate that with teams and individuals, or with applications or systems or business units.“

The tool conducts analyses that Mann said let users understand “which business units are asking for harder software, which development teams make code with more errors.” The application's output includes data on “which teams are able to do faster work, which applications put out more errors, which builds tend to fail more often” and then offers remediation advice.

“It is one thing to report on what's maybe going right or wrong,” Mann said. “It it is an entirely different thing to figure out why it went wrong … applying that to the development cycle and allowing development teams and Scrum leaders to look at the build process and do analytics on build success, build failure, build times or queue lengths.”

Splunk's expertise in IT operations will come into play as the tool interfaces with tools like Puppet to automate provisioning of new resources if required.

Mann said the tool is “a Docker container in a private GitHub repo that contains Splunk Enterprise, Jenkins, SonarQube and a whole bunch of sample data, data normalisation and visualisations.”

While the beta is closed, Mann said a “talk to your local account rep” could result in access.

The tool is significant as it takes Splunk into new markets, but also because is the company's first effort to measure human behaviour.

The company is changing in other ways, too. It's previously sold applications that bundle lots of services. Recently, however, it commenced a tech preview for Project Nova, a free service only limited application logging. Nova is a conscious decision to explore smaller services, rather than suites, while also making Splunk's analysis features an as-a-service offering.

Mann did not say when or if Splunk App for Build Analytics will formally go on sale, but the company is known to time its launches to co-incide with its annual September conference. ®