'Rinse and repeat’: How Nationwide migrated everything in six months

Servers: 9,500. Change requests: 4,500. Downtime: 0

Simplicity, automation and teamwork were the key elements in Nationwide’s on-prem to cloud migration.

For a long time, companies really only had two options for their infrastructure: host everything on-premises, with the associated costs, or move to the cloud, and risk losing control over your data.

Hybrid cloud, blending the best of both worlds, is a third option: an attractive solution when you need standardisation and flexibility while also managing sensitive personal data.

Nationwide, the world’s biggest building society, was in exactly that position. With more than 250 acquisitions in its history, it had “lots of different infrastructure from lots of different manufacturers,” says Paul Walsh, director of IT infrastructure and service delivery.

“[They were] all providing really useful and good services, but clearly made it much harder for us to manage both the technology and the demands of the organisation to move at pace.”

It wasn’t just the organisation, either – customer expectations “are exponentially growing year on year,” forcing a rethink on how Nationwide architects and develops its various platforms.

With some of the old technology unable to transform and modernise, it was time for a change.

After a “pretty comprehensive” review of providers, Nationwide settled on HPE GreenLake as the hybrid cloud solution of choice, operating as an on-premises private cloud – which Paul said offered “a really compelling proposition.”

Part of that was the shift to a pure opex model, which helped fulfil the goal of making Nationwide’s on-premises environments “much more like a public cloud” in terms of scalability and flexibility; but the technical side was just as significant.

“[HPE] had a really compelling end-to-end story on how they brought the various bits of the technology componentry together as a service, and also then not just deliver the service, but be able to provide us with that single commercial rack - and more importantly, that single managed service that sits around it.”

“This needs to be a single team”

Paul was clear early that the migration would need to be a single-team effort. That doesn’t mean only Nationwide or only HPE, but both working together.

"We set out very early on that we wanted this to not be a Nationwide and HPE team, this needs to be a single team: leave your badges at the door and work as a single team. I was really pleased with how both organisations rose to that challenge.”

The combined team built Nationwide's GreenLake platform to sit alongside the existing technology stacks in the datacentre – greatly assisted by both running VMware as a common layer.

“Effectively we stretched the networks from one to the other, used the inbuilt tools within VMware to move the workloads from the old platform to the new platform, and then moved the networks behind them.”

While the work began at a “fairly small scale,” eventually the team migrated 9,500 servers over six months, accommodating about 4,500 individual change requests at the network or infrastructure level. A standardised approach was key.

“The principle we set out very early on was ‘We have a standard way of doing it; once we've proven it once, effectively we can just rinse and repeat it.’ So we then built in automation to allow us to do more of those in parallel, and ultimately then started scaling.”

He adds, “The migration was very successful, and I think it's because we kept it simple and we kept to a very rinse and repeat process that ultimately, we signed off and HPE signed off, so we knew we were working in lockstep all the way through.”

The migration was so smooth that even people in Nationwide “didn’t know we were doing the work.”

What goes where?

Public and private clouds have distinct use cases, especially in a financial business like Nationwide’s, so most apps are hosted on one or the other rather than spanning both.

The nationwide.co.uk website, for example, is “the front door to our organisation,” with heavy digital footfall and variable demand – it works well with public cloud. On the other hand, “some of our more backend core services” are still run in the on-premises environment. The choice depends on characteristics like necessary availability, cost profile and capabilities.

It won’t always be that way, though. According to Paul, Nationwide is looking “very actively” around how to host the most critical workloads across both environments

“[That would] ultimately get the best possible resilience position and make sure we minimise downtime for our services. So, this could be from online channels through to back-office functions and everything that sits in between within a financial organisation like ourselves.”

Paul says the major victory of the project, which was completed in December last year, was in how closely the partners worked together.

“For me that's been the real key driver to the success we've managed to achieve: both organisations having a really clear goal on what good looks like and then coming together in a very joined up way, leaving the badges at the door and really collaborating on driving an outcome... It has been a really stand-out set of deliveries across a really important platform for us.”