https://dvladigital.blog.gov.uk/2014/01/07/migrating-to-digital-platforms-the-dvla/

Migrating to digital platforms @ the DVLA

Several people have asked about our future technology platform and our plans and approach to moving services onto it. So here’s our first update on the subject.

The existing DVLA estate is large and complex, with many of the systems having grown organically over many, many years. These systems have their origins in the paper based world, embodying processes, rules and data constructs which were relevant when records were held on paper and stored in filing cabinets.

These systems are expensive to maintain and required significant amounts of effort to change. It's very clear that these systems will not act as the platform we need as the DVLA offers more and more services as “Digital by Default”.

We'll be moving to a services based architecture, built upon reusable software components, which will allow us to be more agile and more easily adapt to changing user demands. It'll also allow us to adapt and benefit from new technologies and a changing supplier base.

However, given the size of the existing IT estate and the need to continue day to day service there is no way we can move from the old to new technology platforms in a “Big Bang” fashion. We will move services across incrementally over time, “strangling down” the old platforms until we migrate the last elements across.

Effectively we'll re-engineer the business services as we migrate to the new technology platform, but push the data back to the existing IT estate so it continues to be the “system of record”. Then we'll pass a point when the majority of business services are running on the new platform. That'll be the trigger for us to finally move the data across and decommission the old technology.

This approach will allow us to migrate and perform the reengineering that the business services need, while ensuring that risk is kept to a minimum as the old IT estate is kept running until we're satisfied that the new platform operates as required.

The approach is reasonably well known. There’s a good article on it at http://www.martinfowler.com/bliki/StranglerApplication.html.

We'll deliver this in an incremental fashion. As the first services are moved across to the new platform they can be made available. We will avoid massive upfront investment and lengthy lead time before we see results.

It's taken us a lot of coffee and produced more than a few grey hairs to get to this approach! We're now really looking forward to the New Year and seeing it help us quickly bring more services to the digital channel.

5 comments

  1. Comment by David Evans posted on

    Good description of what is planned and good articles to link to. Nice one, Mark.

  2. Comment by Lee Griffiths posted on

    Sounds a sensible approach. I'm sure there will be some snags as UX technology is changing faster than change projects can deliver. As cat enthusiast and computer scientist I'll be watching with interest - hopefully it will make a good case study.

  3. Comment by Lee Griffiths posted on

    Sorry, a few bad autocorrects on the previous post. Cat = car Ps I'd be keen to be a beta tester.

  4. Comment by Stuart Barker posted on

    Service based architecture and reusable components seems a sensible approach. I look forward to reading more.

  5. Comment by Noel Edwards posted on

    This latest post is welcomed, as systems engineering principles are fundamental to the progress of this project. In particular it was interesting to understand how the current systems will be strangled down, as services are moved across to the new platform once fully built and tested.

    This approach, although de-risking DVLA’s day to day business, does however introduce some major technical challenges in terms of the systems integration work required. If understood correctly, the ‘Digital by Default’ remit is for the replacement systems to be re-engineered using solely open source software and database products, deployed on cloud based architecture.

    For the current IT estate to remain as the system of record, then cloud based presumably open source non-sql databases will have to be linked and fully synchronised with existing data centre located Oracle and Adabas databases, such that data can flow freely between the new digital and legacy systems. Given the size, disparity and complexity of the estate, this seems fraught with difficulty, especially with regards performance.

    It would be interesting to learn more about the nature of the cloud provisioning itself, and the functions that the service incorporates, i.e. with regards business data, will backup of this data be part of the cloud service procurement, or will this be retained as a DVLA data centre function? The move to cloud computing would be an interesting topic in itself for a future post, as cloud is such a generic term and it would be good to understand what it means purely within a DVLA context.

    Continuing the theme, will DVLA be moving their whole IT estate to the cloud, including all non-production environments? Presumably the desktop estate will also be moving off MS Windows operating systems onto Linux builds running Ubuntu or similar, with Office and Active Directory being replaced with their open source equivalents? This again is a significant piece of work, which presumably will have to be progressed in parallel with the main system migration, in order to be completed before the end of the current IT outsourcing contract?

    Apologies for the number of questions, but the scope of this transformation is massive, maybe even unprecedented within the public sector, and the timescales for completion appear hugely ambitious. Public sector ICT debacles are all too commonplace unfortunately, and although broadly in agreement with the technical direction being taken, there is nonetheless a major potential for delay and escalating project costs, which as a taxpayer naturally fills me with concern.

    Excellent blog however, highly informative, keep the posts coming and good luck!