Treating large-scale migrations of applications like a factory operation, where tasks are pre-defined, automated and repeatable, is easing the journey to the cloud for large enterprises. Burdened by legacy applications that are formidable both in quantity and complexity, it’s not surprising that corporations are cautious: A McKinsey survey last year found 60 per cent have migrated less than 10 per cent of their workloads to the public cloud.

The prospect of factory-like predictability may bring comfort to big corporations, but it’s vital that they recruit an experienced partner who is adept at dealing with the exceptions that pepper legacy estates. Unless exceptions are dealt with speedily and effectively, they can derail a smooth journey to the cloud, however much automation and standardisation is thrown at the migration.

In a recent migration for an energy company in the United States, the client couldn’t find the source code for an application that controlled access to safety-critical and restricted facilities. Mudasser Zaheer, a digital transformation executive at DXC Technology, explains, “We stepped in and rebuilt the application from scratch”. It’s an extreme example — but it’s not uncommon to have to resolve such complexities and obstacles, he says.

Common problems that need to be fixed by experts in order to be treated as part of a factory migration include: dealing with obsolete versions of middleware or outdated and unsupported operating systems. While tools can automate migration and transformation processes, delivering significant savings in time and money, there’s no single off-the-shelf automation tool — factory migrations need someone who has deep experience.

Transforming underlying database technology in order to simplify a database estate is typically a major piece of any migration. Enterprises have been accumulating diverse technology for the past 15 or 20 years, all of which have different capabilities and features and are embedded into applications. As Zaheer says, “moving them to a modern platform, like the public cloud, is not a trivial matter”.

In particular, many legacy features in RDBMS are not available in native database service from cloud providers. A whole strategy may revolve around simplifying and standardising the estate, if it turns out that the destination platform is not compatible or mature enough to support those features. “In these common factory migration situations, experience counts for a great deal,” says Zaheer.

DXC’s goal is to convert its experience into success, which is defined and measured as a success rate of 99.83 per cent: Out of every 10,000 application it moves, 9,983 move to production and deliver results over and above the service-level objective (SLO). That means only 17 have to be rolled back for one reason or another. Simply moving applications from A to B is not sufficient and can’t be called a successful migration, even if it is achieved using a factory method.

“Success is when applications run, without roll-back, and they are exceeding the service level objectives of the previous legacy platform — they must run better,” Zaheer says. “We ensure that scalability is delivered, including changing application code to reduce [applications’] dependency on deprecated features (features that are no longer there). Ultimately we want to create value for the customer whose customer’s — or employee’s — experience should not be impacted”.