We’ve been saying for a while that deployment automation, self-service catalogs, and modernizing with microservices are key elements for successful DevOps initiatives. But it’s not just us saying that. On Thursday, March 1, three of our customers and an industry analyst got together to prove it.
Here are some highlights:
Chris Condo , Senior Analyst from Forrester Research, presented the roles played by value stream mapping, automating everything, and self-service environments in DevOps success.
Aloisio Rocha , from NetEnt, transformed their ability to on-board new customers and sell into new markets by reducing customer deployment times from 4 hours (with downtime) to less than fifteen minutes with no down time. They can now update their entire customer base in two weeks instead of the 18 months it used to take, a 4000% increase in speed.
Manish Aggarwal , from Intel, shifted quality left with a self-service catalog to manage and automate their 20,000+ test scripts. They now run 1000-fold more tests per day and increased annual releases by 2500%.
Gary McKay , from Somos, future-proofed his company and changed a mainframe operating mindset to an Agile/DevOps culture with microservices. Instead of taking 36 hours and 30-40 people, their releases now take less than an hour with only 5 people.
As a bonus, Aloisio graciously answereed some questions from the audience with details on how they got their 4000% increase in performance (his numbers, not ours).
Q: How long did it take you at NETENT to go from 18 months to < 2weeks?
A: It took about a year and a half with an iterative process. This included evaluation of tools, whiteboarding, learning the tool, building the pipelines, and some refactoring. And also developing trust in the automation pipelines.
Q: Did you stop and pause?
A: No we did not. It just took time to get to a stable, trustworthy and valuable automation pipeline.
Q: Did you get external help?
A: Yes, we had training and some professional services consulting from CloudBees in the beginning to help us understand the tool and how we could use it to fit our specific needs and environment. The rest we did on our own.
Q: What are the hurdles you faced?
A: At first we started the project with a few people that were only partly dedicated to it. So freeing up those people was one hurdle. It had a huge impact when we went for creating a real team with these people, dedicated to the task.
-Getting buy in from stakeholders to help drive the initiative was another one. We were not showing progress and value fast enough because of the way we worked. Our initial approach was to build the automation pipelines as complete as possible before delivering. This led to longer time before we got feedback from users, buy-in from stakeholders, and a loss in momentum.
We changed our approach, and started working with MVP’s (minimum viable pipelines) instead. Which led to faster feedback, which in turn led to more stable and trustworthy automation and faster time to value, which got us buy-in.
Q: Any suggestions for organizations in a similar boat?
A: If you have just started, try to show value or potential early. Prototype it or MVP it. Another key is to get feedback early and consistently - it is critical for building stability and achieving trust in what you are doing.
-It does take time to do something like this, so it is important to manage expectations, but also like I mentioned to show constant improvement and value deliveries. This is key in getting the users and stakeholders to buy in to the future potential, the end goals, and to make it tangible.
-Value delivery analysis and/or value stream mapping are always key tools in these types of initiatives. But keep in mind that they can get very large, complex and sometimes not even understandable depending on the level you are doing them. It makes sense sometimes to split or layer these mappings so that you can get a clearer picture of were to focus efforts and then start to work more efficiently.
So there you have it:
Set expectations early
Start fast with a minimum viable pipeline
Improve the value as you go, and,
Adapt as you learn from your feedback.