Mainframe modernisation offers exciting opportunities for business transformation and growth. But many migration projects stall before they really get going — that’s if they even get going in the first place, of course.
Only 21% of IT leaders recently surveyed by ISG and LzLabs had successfully completed a migration project, with less than half (45%) saying it was a success.
This probably won’t come as a surprise to anyone who’s attempted to modernise an ageing mainframe system. As well as being costly and complex, migrations touch upon some of the most essential applications and transactional data that banks, airlines, insurers and many other companies possess. Mistakes therefore cause big problems – and it’s often IT that gets the blame.
However, sticking with legacy systems that don’t provide flexibility, speed and compatibility with technologies like GenAI is hardly a recipe for success.
Conventional mainframe interfaces struggle to support the real-time speeds needed for modern cloud-native services and technologies, notably, 59% of IT leaders believe this could impact business performance and CX.
Third-party vendor price increases, hardware maintenance costs, technical debt and skills challenges cannot be ignored either.
“If you have a big mainframe system and only two or three people running it, or even one or two people who are close to retirement, that is a huge risk,” says, Dieter Kölbl, vice president of transformation delivery at LzLabs, which transforms existing IT by placing it in a modern computing environment.
IT leaders highlight several key issues that discourage them from migrating, such as a lack of qualified experts to execute modernisation projects (40%) or existing team commitments to other priorities (32%).
Additionally, risk-averse industries like insurance and finance are often reluctant to overhaul well-established systems. Big modernisation projects can also take a long time to deliver results.
“What manager wants to take all the responsibility for a two or three-year program that, at the end of the day, carries a lot of risk?” says Kölbl.
Making the right plan
Some risks are overstated though – or even imagined. Take the lingering belief that the mainframe is the only reliable platform for business-critical workloads, which Kölbl describes as a fairytale. “We, as vendors of migration software, have proven that business-critical [workloads] run very reliably on the cloud,” he says.
Most respondents to the ISG/LzLabs survey are now equally confident that mainframe applications can maintain the same behaviour and performance in the cloud. So how can they overcome migration challenges and unlock the huge potential that modern cloud systems offer?
Firstly, they need to set clear expectations and plan their migration journey meticulously, using in-depth assessments and migration maturity models. Understanding application and data interdependencies is particularly crucial for creating a consistent migration plan that will minimise risk and complexity.
Another key element is selecting the right partners, which should help bridge talent gaps and ensure that security and compliance aren’t compromised.
Of the “7 Rs” of cloud migration strategies, rehosting and re-platforming are the most popular.
For example, 29% of respondents to the ISG/LzLabs survey have rehosted legacy applications, otherwise known as shifting them to the cloud without changes through virtualisation or OS emulation.
Replatforming – recompiling an application to run on x86 or cloud platforms – was the second most popular option, chosen by 24% of respondents.
Kölbl warns against simply converting COBOL code to the modern and widely used Java though. “Companies think that refactoring [restructuring existing source code without changing its external behaviour] is the right thing for them, because then they can get rid of COBOL and have Java…but where they end up is probably not where they really want to be in the future.”
He describes the end result as “JOBOL” – i.e. Java code that retains COBOL syntax – which is “definitely not easy to maintain” and often requires continued knowledge of the legacy language. In other words, you may migrate quickly but fail to reduce the overall complexity of your IT systems, which should always be one of the main goals of a migration project.
The Software Defined Mainframe
LzLabs’ approach, underpinned by its Software Defined Mainframe (SDM), involves gradual migration of mainframe applications to cloud platforms, with APIs facilitating modernisation and integration.
The SDM creates a thin compatibility layer enabling mainframe applications to run on cloud/x86 servers without code changes or data conversion. This binary rehosting approach works even without source code and allows data to remain in its original format, keeping data accessible to modern tools.
Applications can be migrated selectively to the SDM and then gradually modernised. Cutting down the amount of change needed to migrate each application to the cloud also reduces the amount of testing required. Furthermore, other applications can continue running on the mainframe, minimising business disruption and risk.
“You can learn during the migration how the application works, gain trust, and then do bigger chunks step by step until you have reached your final goal, and then you can switch off the mainframe,” says Kölbl. “That’s a better approach because it separates out risks and makes them smaller and more digestible.”
While a single application can undergo modernisation and migration within a few weeks, a well-paced migration plan may take up to three to five years. Kölbl says there are cultural issues that may need to be addressed along the way. “You have to deal with people that have worked with the mainframe for many, many years and…it’s like moving them to a new house…[which can] create all sorts of doubt and uncertainty,” he explains.
He therefore recommends that proper change management is a feature of any SDM migration project. “This can be provided from outside by a third party, or by a big enterprise themselves if they have change managers within the company,” he says.
Once organisations have successfully migrated to the cloud, they no longer face the rising costs associated with mainframe infrastructure, including upgrades, maintenance and annual third-party fees. At the same time, data access is typically improved and data management complexity is substantially reduced.
After migrating to the SDM, organisations can also draw upon a wider pool of IT staff with experience in cloud-native technologies.
The cloud offers multi-regional disaster recovery options to achieve resilience levels beyond what was required in the mainframe era. There are even sustainability benefits due to the reduction in data centre power consumption and heat.
Cutting costs and modernising applications
Many of these benefits are already being enjoyed by LzLabs’ customers. One large North American communications and media company faced rising costs from running business-critical applications – particularly billing – on the mainframe, as well as a dwindling mainframe-related skill pool.
Replacing the mainframe with new hardware and software was ruled out due to inflated implementation costs, as well as the prospect of significant operational support spending in the years ahead. Instead, the company decided to pursue a complete mainframe exit.
IT stakeholders soon learned that Swisscom, a Swiss communications, IT and entertainment firm, had successfully rehosted its core mainframe applications onto the SDM. This prompted them to start a discovery process with LzLabs, to see if they could migrate their legacy billing application, which is over 50 years old, onto the platform.
A key output from this exercise was an actionable plan for the migration of critical applications, plus associated data, onto the SDM while preserving operational service levels.
The company wanted to migrate workloads in stages, maintain interoperability with the mainframe, and perform a final cutover to the cloud with minimal user disruption.
This requirement led to over 5000 test cases being run. That might sound like a lot, but it is actually far lower than usual for mainframe migration projects. Ultimately, 99% of the code was successfully migrated.
Rehosting applications and data to the SDM allowed the company to completely decommission the legacy mainframe platform, creating a more agile and future-proof business based on open technologies. The switch to an internal private cloud is also saving the company millions of dollars in operating, maintenance and support costs each year – a roughly 60% reduction in OpEx costs, in fact.
Strong teamwork between the customer’s internal IT team, LzLabs and the integration partner was a key part of the project’s success. Kölbl believes that having an integration specialist working in conjunction with a solution provider is the “ideal combination” for a migration of this type.
As many IT leaders know, almost every type of technology or CX innovation today happens in the cloud. By migrating mainframe applications to the same environment, organisations can unlock a new world of possibilities, reduce the risks associated with skills shortages and rising costs, and ensure their operations are fit for the future.
So while the best time to modernise your mainframe may have been 20 years ago, the second-best time is now.
For more information on overcoming the legacy tech burden visit LzLabs
Mainframe modernisation offers exciting opportunities for business transformation and growth. But many migration projects stall before they really get going — that's if they even get going in the first place, of course.
Only 21% of IT leaders recently surveyed by ISG and LzLabs had successfully completed a migration project, with less than half (45%) saying it was a success.
This probably won’t come as a surprise to anyone who’s attempted to modernise an ageing mainframe system. As well as being costly and complex, migrations touch upon some of the most essential applications and transactional data that banks, airlines, insurers and many other companies possess. Mistakes therefore cause big problems – and it’s often IT that gets the blame.
