Moore’s Law — you know, the number of transistors in a chip doubles about every two years while the costs are cut in half — seems to have forgotten the mainframe. In the world of mainframes, computing keeps getting more and more expensive, largely because customers haven’t had any other option. Until now.
After years of largely ignoring mainframe applications, cloud is starting to encroach on that last bastion of vendor lock-in. FedEx has announced that by 2024 it will shutter its data centres and the mainframes that go with them to go “all in” on cloud (likely Microsoft Azure, if past purchasing habits continue). In addition to benefits like increased agility, FedEx expects to save $400 million.
More companies will get serious about this sort of decision, which means the cloud vendors also need to get serious about helping move mainframe workloads. What will it really take to convince IT decision-makers? In a very real way, it’s about making the cloud an incredibly boring, safe decision for CIOs.
Boring is as boring does
CIOs might not want to date a boring person, and they might not want to watch a boring movie, but if they’re a CIO, they definitely want to buy boring technology. “Boring” means that pager duty in a company is a perfunctory affair. “Boring” means that things just work. Right or wrong, for decades, the mainframe has been tantamount to boring.
Unfortunately, mainframes are increasingly the wrong kind of boring.
“Boring” can suggest a lack of dynamism, and this is precisely where mainframes find themselves. Even as the world sprints into microservices and everything-as-a-service, mainframes keep enterprises mired in legacy infrastructure. They force enterprises to operate at a relatively glacial speed. They’re boring in the sense of dull and uninteresting, not in the sense of safe.
Some mainframe vendors have tried to convince the market, and their customers, that their mainframe mentality should be an integral part of a cloud future. They’re wrong, and their insistence on pretending that their mainframe revenue is actually cloud revenue only impedes their incentive to change and keeps customers locked into dead-end legacy systems.
As Charles Fitzgerald, a Seattle angel investor and former Microsoft and VMware executive, put it to me, “It has been a really long time since mainframe customers have had any pricing leverage with [mainframe vendors], but the cloud is finally coming for the mainframe. Every customer should have a project with AWS or Azure to move one workload off the mainframe to the cloud.”
Modernisation, one workload at a time
If 50 per cent or more of enterprise data still sits on mainframes, as some estimates suggest, getting off the mainframe won’t be easy — or fast. As FedEx CIO Rob Carter stressed, their journey to cloud hasn’t been overnight, but rather something they’ve been working on “across this decade … [by] eliminating monolithic applications one after the other.”
The key, as Fitzgerald intimates, can be as simple as starting with just one workload, building the muscle and discipline to continue at greater scale. The clock is ticking, however: Employees with relevant skills (e.g., Cobol programming, etc.) are aging out of the workforce.
Meanwhile, every dollar spent on mainframes is a dollar that can’t be spent on business innovation. For FedEx, the delta between cloud and mainframe is $400 million, which doesn’t include the other benefits they’ll glean from moving off mainframes, such as business agility. If cloud is a force multiplier for increased innovation, mainframes are a retardant.
The good news is the big clouds are finally getting serious about mainframe modernisation. Amazon Web Services (AWS), Microsoft, and Google Cloud all have tools and processes to help enterprises migrate mainframe applications to their respective clouds. These should help make a sometimes-arduous process go more easily.
The cloud vendors need something more though, and it’s perhaps less comfortable for AWS or Google Cloud than it is for Microsoft. They need to lobby CIOs the way mainframe vendors have done for decades.
I single out Microsoft because the company has been close with CIOs for eons, while AWS and Google Cloud are still developing their old-school enterprise muscles. AWS has a significant lead here.
That and old-school marketing — coffee mugs and golf games — just might give CIOs the courage to stop marching to the beat of the forced mainframe upgrade cycle and start singing a more innovative cloud tune.