Many companies today have adopted a DevOps approach to software development as they work to meet the challenges of the modern digital business. By breaking down silos between teams, encouraging collaboration, and facilitating agility, the DevOps methodology has quickly established itself as the weapon of choice for companies eager to thrive in the digital age.

But making the move to DevOps isn’t just a matter of melding dev and ops teams and patting yourself on the back. It’s a complex, ongoing process of evolution and re-evaluation. At our recent FutureStack: New York event, New Relic and customer experts opened up about their own experiences, recommending useful tools and offering valuable tips and insights for companies just setting out on their DevOps journey, as well as those that are already far along.

Ushering in stability

In their talk DevOps Takes Practice: How New Relic Evolves Reliability, New Relic VP of Engineering Matthew Flaming and Software Engineer Beth Long suggested that making DevOps work from a site reliability perspective requires “reinvestigating your own relationship to failure.” To explain, they detailed the changes New Relic’s reliability teams implemented following a service disruption several years ago.

See also: 7 Hard Lessons We Learned About Reliability Here at New Relic

“We did get the job done,” Beth said of the effort required to restore service on that occasion, “but the entire process was chaotic and really frustrating.” It also served as a catalyst for major change—specifically, the establishing of the Change Acceptance Board, Gatekeeper, and Upboard, a trio of homegrown internal process tools that “ushered in an era of stability for us that has been really wonderful.”

That stability was tested later during another service disruption. “This time, we had a plan,” Beth said. “We were being transparent both with our customers and internally, and we had the right people and the right processes in place.”

Matthew and Beth’s stressed that waiting for incidents to reveal your software’s problems is not OK. Instead you need to identify your known risks, monitor so you know when issues arise, and mitigate as best you can. “But then you need to start experimenting—poking and prodding at your system so that you can uncover problems before they surprise you.”

Watch Matthew and Beth’s full presentation below:

Operationalize and innovate

Sean Carpenter, senior product manager for New Relic APM, echoed Matthew and Beth’s confession that switching to DevOps isn’t necessarily all smooth sailing. “DevOps is hard,” he said. “It’s really hard. And I like to think of it in terms of two components: practices and culture.”

While culture is a nebulous and nuanced concept that can mean different things to different people—Sean shared some concrete tips that he believes have universal value.

In his talk The APM Growth Curve: Driving DevOps Maturity, Sean emphasized the importance of finding and using the right tools, including New Relic APM. He also broke down the process of embracing DevOps though APM into three key stages: early, stability, and innovation.

In the early stage, he said, you should target existing errors in your application. “Start with the most difficult, the most dangerous, the most damaging thing first, and work your way down the list.” By rooting out and dealing with the slowest transactions, you can begin to establish alerts with dynamic baselines.

In the stability phase, it’s time to “operationalize.” That means integrating APM with your other DevOps tools. The innovation stage is all about customizing your Service Maps and dashboards in APM to drive greater digital success. One technique to drive innovation, Sean said, is to set challenges and then harness the power of DevOps to meet them.

Watch Sean Carpenter’s full presentation below:

Synergy across the toolchain

Similarly, the high expectations of modern tech-savvy customers challenge businesses to stay at the top of their digital games. Today’s customers, explained Tom Kouhsari, lead solution architect at xMatters, are “asking for bigger, better, faster, more … and they want it now. At the same time, they want quality, they want uptime, they want reliability, they want a great user experience.”

In a talk titled How Dealertrack Optimizes the DevOps Toolchain, Tom was joined on stage by Dan Russo, senior manager of production and platform at Dealertrack. For Dan, those high customer expectations make it essential to “turn the vast amount of data that we have at our fingertips into valuable, actionable knowledge.”

Before transitioning to DevOps, Dealertrack’s engineering model was hampered by major interdependencies across disparate teams and a clunky, slow incident-notification process. “We needed a broader suite of tools to provide value to our customers, and we needed technology to help us make that happen,” Dan explained. “We chose New Relic for application performance monitoring, we chose xMatters for targeted communications, we chose Slack for collaboration, and Jira to manage all of our ITSM needs.”

See also: DevOps in Practice at Dealertrack: 11 Things That Help Build a Delivery Culture

After praising the “ton of synergy” he and his colleagues have enjoyed since they started using New Relic, Dan offered three lessons for companies looking to optimize their DevOps toolchain. First, work on your process first, because “tools can’t make up for poor process.” Second, start small: don’t try to integrate all your tools at once. Third, be transparent by making frequent and honest updates a priority.

Watch Tom and Don’s full presentation below:

Azure thing

Like Dealertrack, AgileThought began using New Relic when the company switched to DevOps. That transition also coincided with a move to Microsoft Azure PaaS. In their presentation DevOps with Azure PaaS: Keys for Instrumentation Success, AgileThought Senior DevOps Engineer Art Garcia and Director of ALM Practice Eric Landes described how monitoring with New Relic helped smooth both of those transitions.

Eric explained that AgileThought has so far used New Relic primarily on the dev side, where the company uses .NET, NodeJS, and a database layer, as well as Azure. Getting access to Azure logs can be difficult in certain environments, Art added, a problem that New Relic helped solve. “Out of the box, New Relic works.”

Equally tricky for Art and his team was monitoring WebJobs, a feature of Azure App Service that enables you to run a program or script in the same context as a web app, API app, or mobile app. “New Relic was very responsive, and really came to help us out with a complex issue, and I think helped in a great way.”

These wins have left Eric and Art in no doubt as to the role New Relic will continue to play in AgileThought’s DevOps and Azure PaaS evolution: “We want to make sure that our DevOps processes, our release pipelines, and our builds are pushing everything into New Relic.”

Watch Eric and Art’s full presentation below:

Be sure to check out our ebook Navigating Devops


B.J. Hinshaw is a freelance writer based in Northern California. View posts by .

Interested in writing for New Relic Blog? Send us a pitch!