Jenkins World happened last week in San Francisco. It was good to see familiar faces and meet new friends. The exhibition hall was busy all day long with lively conversations at every turn.
As the baseball season heads into September and the scramble for the playoffs heats up, Yogi Berra comes to mind. Some of his classic sayings apply to our experience at the conference.
Here are a few of our observations from watching the activity at Jenkins World:
People want more
While Jenkins is the de facto tool for continuous integration, and, it appears, the attempted de facto tool for continuous delivery and release automation, the vast majority of attendees are looking for more from their efforts: More control. More visibility. More auditability. Etc. Those demands came from DevOps managers as often as they did from Jenkins administrators.
Apparently, adding more scripts, more instances, and more plug-ins doesn’t solve all the problems of getting high-quality software to market in a timely manner. Being able to reliably automate more steps in the process, immediately knowing the value of software in the pipeline, having a compliance audit be little more than a couple clicks, etc., is still a long way off for many Jenkins shops.
People want less
In theory, the whole point of automating development and operations is to eliminate bottlenecks, errors, etc. Yet, based on comments and questions by attendees, the practice hasn’t been all that successful.
One DevOps manager from a shop with multiple Jenkins masters, about 20 development teams and “lots” of scripts admitted that managing the whole affair was a major pain in his Anterior Spinal Section. He wanted less pain, fewer errors, and less time chasing up problems. In his shop, jobs that fail are automatically restarted ad infinitum and get logged. Reports are handed out at the weekly stand up meetings with directives to “go fix your broken Stuff Hindering Impending Triumph.” He wasn’t alone.
“Doing DevOps” is still a poorly defined concept
One attendee, wearing a rather bewildered look, came by and admitted that she was there to learn as much as possible and where to start. It seems her Upper Management had decreed that the company needed to be “doing DevOps” in the next 90 days. With no experience and no definition of what “doing DevOps” really means, it’s hard not to invoke the above classic Yogi-ism. The additional problem with implementing automation technology without a clear idea of goals and metrics is not only will you wind up someplace else, you’ll get where you don’t want to be faster than ever before.
Even though it’s late 2017, it’s clear that the DevOps game is far from over. There are still way too many islands of automation, too many manual processes, too little understanding of how to make it work optimally, and too little visibility into the value of what’s in the pipeline. The inevitability of digital transformation will continue to force the slow or the inflexible – both businesses and tool vendors – out of the market.
Latest posts by Tim Johnson (see all)
- AI, DevOps, and Employment Law: Thinking Ahead - December 10, 2018
- Loving Your Pipeline’s Middle Child: 60X Test Acceleration - December 7, 2018
- More Speed with ElectricAccelerator – Part 2 - November 28, 2018