How a new focus on inter-team relations can refresh your DevOps culture
Thu 4 Jan 2018 | Alex Stanhope
Ahead of his session at the forthcoming Cloud Expo Europe 2018, Alex Stanhope, Head of DevOps at the Department of Work and Pensions (DWP) discusses how his team is working to improve quality and efficiency by streamlining interactions.
While it may seem that work within our engineering team is forever changing, the fundamental principles of what we are trying to achieve are fairly constant. We want to build great citizen-facing web services that are focused on our user needs.
What does change are the tools and technologies we use to achieve that. While three years ago all the basic elements were there – cloud platforms, provisioning tools, configuration management tools, web frameworks, monitoring, application performance management etc. – they’ve now grown in their accessibility, sophistication and maturity.
To have an iterative and proactive cloud strategy is essential, but so is the DevOps culture that trumps it
It is easier to provision or decommission hundreds of server instances in just a few seconds, set up or inspect security groups, stand up blue-green environments or react to ever more complex cyberattacks.
The changing face of technology brings with it some challenges – patch management is as important as it’s always been – but predominantly opportunities for more dynamic, more highly optimised and better value cloud infrastructure.
To have an iterative and proactive cloud strategy is essential, but so is the DevOps culture that trumps it. If teams of developers and operations staff can work together seamlessly, we have got everything we need to make the most of the next technological wave.
Prioritising team relations
The security of the data we hold is paramount. Second only to that is the availability of the services we provide. Every role plays a part;
- The requirements that business analysts create have to be checked to make sure they’re comprehensive, compliant and robust
- The designs our architects produce have to be governed, assured and peer reviewed
- The implementations that our engineers code have to be thoroughly interrogated by penetration tests and static code analysis
At every stage, risk is appraised and mitigations planned. Doing all that is hard, but we also need to be agile, learning and reactive to the user needs and service context. Doing it fast is harder.
Team communication is thus a priority as we start to increase the cadence of all the connected activities. Intra-team communications can be managed by good agile ceremonies (regular stand-ups, retrospectives, planning) but inter-team relations is a tougher nut to crack. After all, there are only so many people you can fit into a meeting room or team space.
Beyond the healthy activities (meeting discipline, prioritised and well-structured email, chat channels, PM tools etc.), the biggest guide I have found is lean. We looked at the complexity and regularity of the interaction between our teams and identified patterns. A lot of time was being spent customising or rendering bespoke what was essentially the same activity over and over again.
The underpinning principle of collaboration and knowledge sharing will be around forever
We have concentrated on the interfaces between teams: scrapped redundant or out-dated forms, avoided ground-up discovery sessions and where possible reduced the team-team interaction to a well-defined, versioned API. This API approach can’t be applied in all situations, but it has meant that each collaborating team has developed a set of services they offer.
As the services have got more mature, the quality and efficiency of the interaction has improved: everyone gets time back in their diary, which leads to more efficiencies!
Looking to the years ahead
In the future, I think we will give DevOps another label, but the essence of what we’re trying to do is centuries old. I’m sure great British engineers through history have faced similar challenges. When Brunel was building Bristol’s Clifton Suspension Bridge, I reckon the assembled group of collaborating engineers and technical specialists will have had to work out a lean efficient operating model to communicate the vision, react to what the teams are learning on the ground, design and implement novel solutions to unpredicted problems.
In five years’ time, DevOps will be gone or dated, but the underpinning principle of collaboration and knowledge sharing will be around forever.
What’s next? RobOps perhaps!
Alex Stanhope will be speaking at the forthcoming Cloud Expo Europe, London, 21st and 22nd March 2018 at ExCeL London. To hear from Stanhope and other DevOps experts from around the world, register today for your FREE ticket.