From 0 to DevOps in 80 Days
In his speech, Bernd Greifender shares how Dynatrace has come from a meagre 2 releases a year and a product used on premises (2011) to a SaaS model with 26 yearly releases and 170 daily deployments (2016).
Dynatrace is a monitoring solution for distributed applications, it’s a reliable performance management tool that can be used for up to 120 000 hosts. In his speech Bernd Greifender shares how Dynatrace has come from a meagre 2 releases a year and a product used on premises (2011) to a SaaS model with 26 yearly releases and 170 daily deployments (2016).
This required a lot of intra-company changes - monitoring software construction - testing, integration, releasing, introducing more automation, setting up specific infrastructures and a new approach to team and people management.
Yes, they’ve mastered DevOps. But how exactly did they do it? How do you manage the engineers, software ninjas and product managers and create a smooth delivery process that works seamlessly and actually brings results?
A SaaS model needs continuous integration, continuous delivery, deployment, and feedback, and the first thing to master was continuous integration. But to get to full automation, DevOps (or even NoOps) and make it from code to production in one hour (yes!) what actually was needed was a bigger cultural and organizational change. If you’d like to go from O to DevOps with your team the most important part is managing engineers – making them responsible for about everything to the customer, encouraging pro-active attitudes, and giving real-time feedback as products are used.
One of the first things that Dynatrace did is made sure engineers did the Q/A testing whereas all demos were run in the real environment. There was no more safety on your own machine, only real life circumstances - this shifted focus, made the engineers aware of various component dependencies and as a result led to better performance.
A mind shift and culture change among the workforce was also important and Dynatrace used quite an atypical approach to support this – a live visualization of the built pipeline found its way into the staff cafeteria. If you’re curious about the details just check here: https://dynatrace.github.io/ufo/.
One of the pains of the operations team was the responsibility to maintain products they haven’t built and thus didn’t have much insider knowledge, this led to problems. Introduced solution? Chief software architects got similar power as Product Managers with regards to the final product and they had to agree on unanimous solutions.
Also, as the company needed to operate 24/7 in order to achieve its goals focus was made on total automation. A closer look at the run books led to automating runbook entries – as a result, one entry was left.
But all of this would never be possible without the management supporting this vision. Just the orchestration team was 7 people’s work over a course of two years. As this was an upfront investment it was necessary that management understood the vision and reasons behind these expenses.
If you’d like to speed up your application lifecycle watch the full video here:
Bernd Greifender is the CTO of Dynatrace, an entrepreneur, innovator and software performance expert in one. A multitude of specialties, years of experience with startups result in plenty of knowledge which he eagerly shares at tech and entrepreneurial events.
Tags: DevOps, Dynatrace, total automation, from code to production in one hour, speed up application lifecycle, automating run book entries, set up infrastructure, monitoring software construction,
HAVE AN IDEA FOR CONTENT?
Contact the editorial team at:firstname.lastname@example.org