4 Management Strategies to Turn Around Any Software Development Project
A digital operations manager for Allrecipes.com explains how he navigated a particularly tricky cloud migration.
On Wednesday, November 25, 2015, Allrecipes went dark. A site outage is never good, but this one just happened to hit the day before Thanksgiving — by far the busiest time of the year for the do-it-yourselfer’s digital guide to kitchen kickassery. Further complicating matters: The site’s deployment team was in the early stages of a massive effort to migrate to the cloud.
The next 12 months brought with it a series setbacks that threatened to sink the migration project, including over-empowered consultants, several architectural do-overs, and hundreds of learning experiences that the team earned the hard way. Oh, and if that weren’t enough, the director of operations left, followed by the CTO. Then, just a week before the migration was scheduled to take place, the site was hit by a massive DDoS attack, forcing the six-person team to flip the switch before the site was truly ready.
Leading the transition fell to Merritt Bettineski, who moved up from an individual contributor role six months into the project. Despite the challenges, though, he and his team managed to pull off the migration and do a ton of tuning and cache optimization in the cloud. And by Thanksgiving 2016 the site was in such good shape — thanks in part to an all-nighter Bettineski and one of his systems engineers pulled the night before — that it held up through what turned out to be Allrecipe’s heaviest day of traffic ever.
Throughout his six-month trial by fire, Bettineski had a handful of epiphanies that led to the following useful tips for weathering any major project that seems to be teetering on the edge of collapse.
Solicit input
When Bettineski assumed the role of digital operations manager, he inherited a talented team that had been demoralized by migration challenges and a leadership vacuum. To get them — and the project — back on track he consulted with each member to get his or her assessment of various processes and how they should be handled. “Soliciting input lets people know their opinions are valued and respected,” Bettineski says. “From there it’s easy to give people ownership. And once people have ownership, they have skin in the game.”
Stop trying to assign blame
It’s a fact of software development that in any complex, distributed system designed and managed by multiple teams, things are going to go wrong. You don’t have to like it, but that doesn’t mean you need to waste time trying to find and chew out the person or team responsible, either. Instead, work to set up rules or preconditions to prevent the issue from happening again. “My attitude is, I assume you’re doing excellent work 99 percent of the time,” Bettineski says. “If it was a mistake, then it’s correctible. Let’s move on and move forward.”
Ditch deadlines for large projects
Like everyone else, Bettineski’s team tracks work items. But unlike a lot of teams, they don’t worry about estimating how long it will take to complete them. They pour the energy required to do that into breaking down large, seemingly impenetrable projects to the “atomic” level, creating actionable items that aren’t too intimidating to work on. And from there, things fall into place. “It’s hard to explain why, but as soon as you start taking away some of the deadlines and expectations, work just starts happening,” Bettineski says.
Encourage partnering to solve problems
When work piles up, it’s easy to see multiple people working on the same task and start worrying about inefficiencies. Don’t. “When I see two people working on a problem I don’t think, ‘They could each be working on a different problem,’” Bettineski says. “If they get that one big problem knocked out twice as fast and they can each go back to their individual projects, that’s a win to me.”
By the way, Thanksgiving 2017 was, as Bettineski puts it, “a dream.” Because he and his team had put in the work to smooth out the kinks leading up to the success of Thanksgiving 2016, they were able to coast into the following year’s holiday weekend. Well, almost. “We did a lot of forecasting and took a lot of precautions, like turning off auto-scaling and setting up intentional scaling,” Bettineski says. “As it turns out, we probably could have gotten away with letting things auto-scale and not doing anything special. But it felt good to be ready.”
Want to ship software faster to production? With its weeklong Navigator program, sodo will bring technical subject matter experts to you to help you identify your team’s skills and set goals for a software delivery transformation. https://l.shipondayone.com/nav.html
