How the UK Department for Work and Pensions cut deployment times to 15 minutes

The UK’s Department for Work and Pensions says it has been able to reduce the deployment times for its cloud-native services from hours down to 15 minutes, by automating key deployment steps and running on a common container orchestration platform.

Formed in 2001, the DWP is responsible for a variety of public services in the UK, including various welfare, pension, and child maintenance payment schemes that serve as many as 22 million claimants a year.

Demand for its services soared during the pandemic as unemployment rose, driving a desire across the various feature development teams to speed up deployment and respond to that changing demand.

Establishing a common cloud-native platform

Starting 18 months ago, DWP Digital, which employs 5,000 people and is responsible for the department’s technology estate, reached out to Red Hat to explore how a common layer of its OpenShift container management platform could enable all engineers to consistently deploy their microservices to a single platform, regardless of the underlying infrastructure.

“Time and time again we saw teams wanting to adopt cloud-native ways of working, but we wanted Red Hat to help us collaborate on a common platform and open source ways of working,” Dean Clark, technical lead at DWP Digital told InfoWorld.

Historically, the DWP has operated a technology estate via autonomous teams running their applications both on-premises and on a mix of Amazon Web Services (AWS), Google Cloud Platform, and Microsoft Azure public cloud platforms.

That included a bunch of self-managed Kubernetes environments. “Everyone was standing up an environment they would manage, even if it was a managed service from AWS,” Clark said. In response, DWP Digital started looking into establishing a more common platform, built on OpenShift and managed by what they call a “cross-cutting team” of experts.

“There was a lot of cognitive load on feature teams to build pipelines and artifacts,” Jacqui Leggetter, head of integration for DWP Digital told InfoWorld. “We recognized that we were repeating things and could ease that cognitive load and free them up to deliver better citizen outcomes.”

However, handing down a platform from on high would never work across such a wide variety of previously autonomous teams. “People will throw rocks,” Leggetter said. “You have to co-create and the central team worked really hard on collaboration. The key to the success of that common platform is to create a community around the central team.”

Automated playbooks speed delivery

On top of that, the DWP wanted to use the Ansible automation tool to write repeatable playbooks for developers, “to accelerate our path to delivery, be consistent, and remove the repetitive tasks we were seeing for feature teams,” Leggetter said.

By leaning heavily into building event-driven APIs across the department, combined with these playbooks, DWP has been able to create a “consistent user interface with better discoverability across departments to reuse modules and playbooks,” Clark said.

By automating tasks such as environment builds and standing up monitoring and alerting, DWP developers are now able to more quickly build and update their microservices.

For example, when the UK government made changes to the Universal Credit benefit last year, DWP Digital was able to build and launch a new API for automated identity verification in three weeks, as opposed to what would have been five or six months’ worth of work under the old model.

Building a common platform

While the DWP as a whole is still not close to running all of its applications in containers and deployed via a common internal developer platform, that is very much the goal.

“They are not all in that place, but we now have the reference architecture to take us there,” Leggetter said. As part of that, any new service must be built as a microservice, using event-driven APIs and deployed as a container on OpenShift, with existing systems moving to that architecture at their own pace.

“We have a long way to go with the number of systems, including wrangling some of our heritage systems—including those written in COBOL—where opening up the data within those platforms into APIs will help us to catch up with the new services,” Leggetter said.

The DWP accelerates its app modernization

One of those existing systems is the COBOL application supporting the UK’s Jobseekers allowance, which was migrated from a mainframe environment via a “like to better” conversion to the object-oriented Micro Focus COBOL in 2020.

“That was an intense time to do an application migration for a benefits service that was starting to see an avalanche of claims because of how Covid was hitting the country,” Mark Bell, the virtual machine environment replacement (VME-R) program lead at the DWP, told InfoWorld last year.

Where updates to the old monolithic COBOL systems could only be deployed once or twice a year, today smaller, more regular changes can be made by the DWP developer team. Those developers can also start to experiment in a dev/test environment on Amazon Web Services, building out a set of reusable APIs to expose key data sources and pushing changes through a CI/CD pipeline.

DWP Digital says that its success in combining containerization with automated playbooks has led to other teams across the UK’s Civil Service to become curious about using these types of cloud-native technologies themselves.