Department for Work and Pensions

Converting 25 million LOC and 11 billion database records

Following a series of Proof of Concepts, the team selected a re-platforming approach utilizing Visual COBOL and COBOL Server. Amazon Web Services (AWS) was leveraged to host the development environment. Application code conversion specialist Advanced was engaged to work closely with DWP Digital. With 25 million application Lines of Code (LOC) and nearly 11 billion data rows to convert, DWP decided on a phased implementation which enabled them to learn from tackling each system in turn and apply that experience to the next system. This was not just a technical undertaking. The new platform would enable increased DevOps maturity and embrace ‘SCALED’ agile delivery using rapid deployment and Continuous Integration/ Continuous Delivery (CI/CD) capabilities. This was a significant culture shift for the development and support teams and required cross planning and training alongside the modernization effort.

Using Visual COBOL, the business applications’ old COBOL code was converted to the more modern COBOL version. The first solution to be moved over was the relatively small Housing Benefit application, paying out £1.5 billion per month. “Thanks to all the preparation, the move to the new environment went seamlessly. We were delighted to discover that the nightly batch processing was reduced from 90 minutes to less than 15 minutes,” says Bell.

Moving through the solutions, including War Pensions, Industrial Injuries, and Social Fund, the team streamlined its processes by introducing automated testing and scripted business scenarios. This enabled comprehensive comparative testing on transitional activities between old and new systems to assure that the financial output and the number of transactions were 100 percent accurate. Because the individual benefit systems integrate closely, i.e. an award of one benefit might impact another, every change needed to be tested and retested, supporting cross platform testing between the different systems.

Flexibility in time to respond to COVID-19 crisis

The Job Seekers Allowance Payment System (JSAPS) was the largest service with integration to 54 separate database instances and 6.4 billion rows of data. “The timing to move the Job Seekers Allowance system coincided with the start of the COVID-19 pandemic,” says Bell. “We had planned a 4-day implementation period over Easter 2020 when we would do this. With our teams now all working remotely, we worried about the feasibility of our plan. However, COVID-19 also gave us a compelling reason to prioritize this effort. We feared that this service would need to support a lot of additional users, as DWP increased its response to support its UK citizens because of the pandemic, and had concerns over the old platform’s ability to manage this.”

It took just 24 hours to extract and convert all JSAPS data to the new Oracle database instance, allowing the data to be exploited more easily as part of the digital agenda. Final go-live in the new environment was achieved in three days—a day ahead of schedule—and with no disruption to citizens or DWP users, to Bell’s great relief: “Since then, the service has been incredibly efficient, with batch run improvements of 60 percent. Thanks to the flexible and agile new environment we could quickly make government-mandated policy changes to allow for COVID-19, such as increasing our capacity for additional users, and supporting quicker benefit processing changes. In February 2021, the program won the Civil Service Award for Project Delivery Excellence for the JSAPS delivery.

Disability Living Allowance, Income Support, and Pension Service solutions soon followed in the modernization project, and within a two-year timeframe all critical DWP solutions were converted by leveraging Visual COBOL. The ability to deploy changes into the solutions has accelerated by at least 75 percent with an automated build and deployment framework. Bell explains what this means in practice: “Leveraging Visual COBOL we created a CI/CD and container pipeline approach in AWS to give us rapid deployment capability.

Instead of two major releases each year, over the last 12 months we deployed 800 times. Code changes that would have taken us many weeks in the past are now completed and deployed within a couple of days. Thanks to this flexibility, we can do mid-week changes, rather than booking out entire weekends for system maintenance. This means we can respond to citizen requests for extended opening hours. And in some instances we have also managed to completely clear our backlog of change requests.”

Leveraging Visual COBOL we created a CI/CD and container pipeline approach in AWS to give us rapid deployment capability. Instead of two major releases each year, over the last 12 months we deployed 800 times.

60% Faster batch processing and nearly £150 million cost savings

Streamlining batch processing has changed the way in which the service operates, and the team works. Batch processing used to take an entire night with staff on hand to manage at least 3,000 intervention alerts each month. Batch processing is now much more robust and up to 60 percent faster, allowing for patches, fixes, and small changes to deploy during the same window of opportunity. On JSAPS alone, the batch intervention rate has dropped by 96.75 percent to just 100 intervention alerts each month, which is expected to decrease further. “Modernizing our back end infrastructure, re-engineering our batch processing, and converting our code to modern COBOL offer significantly greater stability and give us business and technical opportunities that we look forward to exploiting further,” comments Bell.

“This re-platforming project with Micro Focus (now OpenText) and Advanced was always about creating a stable and secure system. Of course, achieving nearly £150 million Net Present Value (NPV) in the business case is very welcome,” according to Bell. “The use of AWS gives us a modern development environment with fast and efficient release management. Having such a modern infrastructure to support these key UK services gives us the opportunity to offer apprenticeships which will attract new talent into our organization to support our ‘Digital with Purpose’ transformation agenda.”

He concludes: “The size and scale of this project made it technically complex and very challenging to achieve. The success and incredible results are thanks to exceptional collaboration between DWP Digital’s Service Planning & Change Delivery team, our partner Advanced, and Micro Focus (now OpenText). The effort has been recognized with industry awards and very favorable audit results. We have delivered a new and agile way of working with a continuous improvement approach. Our initial ‘like for like’ mandate has turned into a ‘like for much better’ result and we couldn’t be more pleased.”