Peter Sayer
Executive Editor, News

In transition: How Kyndryl’s CIO weaned the company off IBM’s systems

Case Study
Nov 09, 20237 mins
Application ManagementCIOEnterprise Architecture

When IBM spun out its managed infrastructure services business as Kyndryl in 2021, the new company had two years to transition to its own IT systems. It also chose to use that time to undertake a major transformation.

Kyndryl logo on smartphone
Credit: T. Schneider / Shutterstock

For Kyndryl CIO Michael Bradshaw, the clock started ticking in November 2021 when the former managed infrastructure services division of IBM was spun out as a separate entity and given two years to disentangle its IT systems from IBM’s.

“We had a 24-month transition services agreement,” he says.

With that deadline came a dilemma: “Do we just focus on transition, then start thinking about what transformation looks like,” he says, “or do we use this as an opportunity and compress the cycle to drive transition and transformation, using the platforms as the driver for that.”

With the rest of the Kyndryl leadership team, Bradshaw opted for the latter.

As a result, some other processes were compressed too, including the choice of the key platforms on which Kyndryl’s new infrastructure would be built.

“What typically happens is, you get these long drawn-out RFP processes and all the balancing and waiting, and the reality is, it doesn’t really matter,” he says.

Bradshaw recounts how Kyndryl’s HR management platfrom was selected.

“I sat with my CHRO in her first week on the job, and said, ‘If I asked you to choose between SuccessFactors and Workday, what would you choose?’ And she said, ‘Well, the past two jobs I’ve implemented Workday, so my answer would be Workday.’ And that was pretty much our decision.”

For the ERP, Kyndryl went with SAP, the same platform used by IBM, although it’s possible that Bradshaw’s 30 years of experience with SAP weighed as heavily as IBM’s history with it.

He defends what might at first seem like a frivolous selection process, pointing to the predictability of traditional methods and their failure to facilitate change.

“We didn’t do fit-gap analysis workshops because 95% of the time, the solution looks exactly like what the teams have today,” he says. “We weren’t going to go down that path.”

That may seem odd, given that two key platform choices were dictated, at least in part, by their familiarity. But there’s more to it than that.

“One vendor versus the other isn’t really that important,” Bradshaw says. “It’s about living in that platform.”

In any case, he says, only 13% of Kyndryl’s revenue flowed through those legacy SAP systems in IBM, with the rest going through custom mainframe applications. Kyndryl was not given a clone of those IBM systems as part of the transition services agreement, he says, so they had to start afresh with the SAP installation.

Kyndryl is now running a clean version of S/4HANA Cloud private edition on Microsoft Azure.

That’s had its advantages: “We’ve been able to leverage these platforms to do so much, more rapidly, and with much more agility than we would have if we had taken the more brownfield approach,” he says.

Living in the platform

The move to a greenfield SAP system illustrates what Bradshaw means by “living in the platform.” Rather than customize a system to fit exactly what had gone before, he wants to get the most out of a platform’s innate capabilities.

“It’s what is differentiating for the business that really matters,” he says. “For those non-differentiating capabilities, you shouldn’t spend a lot of energy; you should do the base configuration and then shift the business toward the different solutions.”

Another example: Kyndryl — or the IBM business unit that eventually became Kyndryl — already used Workday before Bradshaw arrived, but only in a limited way, alongside dozens of other HR applications. “The choice to expand on Workday was to double down and make that the platform, collapsing all of those 60-plus different applications and leveraging that platform,” he says.

For some aspects of those migrations, including configuration and test management, he brought in external systems integrators: KPMG to help with Workday, and Deloitte for SAP applications such as Ariba and Fieldglass. For other aspects, such as setting up application environments in Azure, he had help from Kyndryl’s own consulting staff.

One advantage Bradshaw has that CIOs in other kinds of business don’t is he can turn learning experiences from costs into profit centers for Kyndryl if a task turns out to be something the consulting team can sell to other companies as a service.

That was the case with the migration of quote-to-cash and financials data from IBM’s legacy SAP system to Kyndryl’s greenfield instance.

To develop the tools to do that, Kyndryl partnered with SAP, and now offers its customers a data migration service using those tools.

Transferrable mainframe skills

Kyndryl’s old mainframe applications have been migrated, too.

“We now have no mainframe applications,” Bradshaw says. “We’re completely running in Azure or are software-as-a-service based.”

Other CIOs are moving their applications off mainframes as well, perhaps in part because they’re concerned they may not be able to find skilled employees to maintain them in the future. This can become a self-fulfilling prophecy as new IT graduates shy away from mainframes and turn to cloud technologies instead to avoid finding themselves in a career dead-end.

But Bradshaw sees a way forward for both existing mainframe experts and those learning about them today.

“Folks who have a mainframe background are perfectly suited to manage hyperscaler environments and even across multiple hyperscale environments,” he says. “The core disciplines you use to run and manage a mainframe well are the same disciplines you have to exercise to manage those hyperscaler environments.”

That equivalence, however, didn’t become apparent to Bradshaw and his colleagues as early as it could have.

“We could have brought more people along on that journey,” he says, a move that might have enabled Kyndryl to complete some of its application migrations sooner.

The surprise for Bradshaw was that staff were typically able to pick up the skills necessary to transfer from mainframe to hyperscaler management in four to six months, with the best making the switch in just two or three months.

“I didn’t think we could go up the scale that quickly,” he says.

Support calls

Bradshaw has one other regret, besides overlooking the transferrable skills of his mainframe staff: not kicking off the organizational change management process sooner to better communicate the notion of living in the platform.

As it was, he communicated through the business leaders, with each of them relaying the message in a language their teams would understand. The focus was on the most important use cases, and on making the change without disrupting customers’ activities.

The process was successful in that staff knew where to go for help, he says, noting that it’s not enough to give them 20 hours of classroom training and then leave them to it: you have to give them a help line to call and let them know they’ve got support.

Even if they call to say, “’I know we covered this and I feel really silly to have to ask,’” he says, “Don’t make them feel silly.”