Beyond the moral support, the executive management coaching and the tips on how to build DevOps momentum in the enterprise environment, last week’s DevOps Enterprise Summit 2015 offered a boatload of excellent fodder for those seeking to prove the value of DevOps patterns once they got back home.
DOES wasn’t just about cheerleading—alongside lessons on the soft, squishy elements of transforming IT, nearly every speaker offered up stone cold numbers to back up why those transformations make business sense. The performance metrics shared by speakers were impressive and could go a long way toward swaying skeptical stakeholders. Here are some of the highlights.
Ticketmaster CTO Jody Mulkey offered a ton of excellent examples of how an organization-wide commitment to DevOps has fueled innovation and stability at his company. For example, he described how what Ticketmaster calls the ‘Boom Team’ has used a completely integrated team of devs, operations, QA and other staff of about 20 to shift its Ticketmaster.com property—which accounts for 40 percent of the company’s revenue—from a duct-tape and bubblegum affair with 2000 ModPerl at its base into a modernized system continuously delivered by that team. Similarly, its technical operations team has embraced the DevOps ethos in order to deliver what it calls ‘support at the edge’ for WebOps incidents. The success of this approach has resulted in:
- 90 percent reduction in mean time to repair (MTTR)
- MTTR from 47 minutes down to 3.8 minutes
“This was a phenomenal change,” Mulkey says. “The business went crazy over this. These are measurable outcomes that have really helped demonstrate that we are making progress.”
Target brought forward speakers to discuss the retailer’s innovation in a number of key areas. Heather Mickman, director at Target, talked about how her API team worked hard to not only easily expose product and store information via its APIs, but to also challenge itself in the way it was delivering its work.
“We wanted to deliver new capabilities in days and not months,” she explains, saying the team needed to be sure that it could scale its APIs because it new that they would be highly reused—both internally and by partners. Target has:
- more than 90 APIs
- monthly volume of calls to these APIs have grown in one year from 1.5 billion to 17 billion
- deploy 80x per week across stage and production enviornments
- all that, with fewer than 10 incidents per month
- this volume helped support in 2014 a 42 percent increase in guest traffic and 280,000 orders fulfilled on Black Friday alone
HP CIO Ralph Loura says HP didn’t go for the big bang approach, but through careful pilots and the use of communication afforded by ChatOps infrastructure, his team has been able to drag HP IT slowly into the DevOps fray. That’s a good thing, considering the massive scale of the organization:
- 7,000 developer and operations staff
- handling over 20,000 change requests per year
- supporting 1,400 apps
- and 1000s of builds per day
After four years of hard work to move toward DevOps and continuous delivery, CapitalOne Product Manager Tapabrata Pal had a number of impressive statistics to share on its its progress:
- Code commits have moved from random occurrence to hundreds per day for any given product team
- Integration has moved from 1 per day to every 15 minutes
- QA performance has shifted from monthly to four times per day
- Production releases have moved from quarterly to one per sprint
- Defects have been reduced by approximately 60 percent
- Feature delivery rate has increased by approximately 50 percent
Paula Thrasher, application delivery lead with CSC, stepped up in front of the DOES audience to show them how it is possible to battle bureaucracy at any organization—coming from the government sector, she has first hand knowledge of what it takes. In one example she shared, her organization helped a client move its cyclew time from automating deployments to staging:
- from 12 hours down to just 12 minutes and 30 seconds.
IBM’s Angel Diaz, VP of cloud architecture and technology, talked about how enterprises can still innovate at the pace of startups while delivering at scale. The thing with IBM is that it puts its money where its mouth is when it comes to DevOps—the company has doubled down to not only offer a number of products viable for the continuous delivery pipeline but to also completely transform its own business through DevOps practices. Some of the results it has seen by shifting its 2,000-plus developers and 100-plus squads in the Cloud Services division to DevOps:
- 300 percent increase in “pure coding time” for its Cloud Services developers
- reduction of deployment time from 32 hours down at first to 6 hours and now to minutes
- an increase of release frequency from months to every three weeks.
Disney’s Jason Cox, director of systems engineering, got the crowd pumped up with a slick mash-up video of Star devWars and DevOps themes (not to mention a private showing of the most recent Star Wars trailer at the end of his talk!). But he didn’t favor style over substance. In his talk he described how much success Disney has grabbed up through DevOps, including:
- 75 percent reduction manual steps
- reduction in deployment time for large online/digital experience systems from 3 hours to 20 minutes
- ability to provision 300-700 srvers within 15 to 30 minutes