By Owen Wheatley, partner, and Sowmiya Bakthavatchalam, senior lead analyst, ISG
Mainframes will remain in large financial institutions for some time to come, but many banks are grappling with whether to modernize or outsource them. The challenges posed by mainframes are significantly shaping enterprise digitization strategies and trends in the financial services industry.
Mission-critical applications in some of the world’s largest banking and financial services (BFS) enterprises still run on mainframe computers. Their sheer computing power and robust security features have made mainframes a trusted technology for BFS enterprises that want to scale their operations and ensure their data remains secure without major process breakdowns. The transactional integrity and pervasive encryption that mainframes offer remains unparalleled, and they easily support millions of daily banking transactions and manage critical workflows.
Despite the advent of technologically advanced cloud alternatives, the global mainframe market is still projected to witness substantial growth in the coming years. Mainframe and cloud deployments will continue to co-exist as enterprises harness the strengths of both worlds to maximize value and scale. So why do so many anticipate the demise of mainframes?
Navigating Mainframe Challenges
The popular perception of mainframes is that is that they are equivalent to heavy machinery in a factory. Mainframes were designed to manage high volumes of data in financial institutions that need to perform large-scale transaction processing, support several thousand users and application programs, handle terabytes of information, and leverage large-bandwidth communications. And mainframes do this well. The challenge today is the cost and skill needed to maintain them.
Many banks and other financial institutions are trying – where practical – to move away from the mainframe to address the widening legacy systems skills gap, the increasing cost of running software on the mainframe, and the perceived disadvantage of missing out on emerging cloud technologies.
However, moving away from mainframes has been more easily said than done for most large, established banks, due to the cost of migration and the significant risks associated with such moves – think heart-and-lungs transplant. In addition, most technical experts will agree that mainframes still offer the most secure and effective platform for the large volume of workloads and IO-bound transaction applications that are typical of financial institutions.
An emerging workaround for the mainframe migration conundrum is to develop a cloud-native infrastructure and use the “strangler pattern” method to incrementally replace legacy applications with new ones as part of a broader dual-core strategy. Alternatively, many large banks leverage third-party outsourcing to drive mainframe optimization and address the resource crunch.
Why It Is Still Not the End for Mainframes
The pandemic ushered in seismic changes in the banking and financial services sector, forcing firms to reconsider their reliance on five-decade-old mainframe technology to save costs and adapt to increasing customer demands. Demand for more modern technology is tempered by the realities of entrenched mainframe usage: Mainframes handle billions of ATM, credit card and trading transactions every year, and more than 200 billion lines of COBOL code remain in production, a number that’s increasing annually. This is precisely why the leader in the mainframe space, IBM’s Z platform—which in 2020 was used by 44 of the top 50 banks—has seen multiple reinventions, such as the inclusion of cloud-native development capabilities and enhanced processing power. But are these adaptations enough to keep pace with emerging, innovative technologies and rapid business growth?
The spiralling cost to maintain a mainframe environment is compounded by the rapidly dwindling talent pool. In one notable example, the governor of the state of New Jersey, Phil Murphy, publicly pleaded for more COBOL programmers in April 2020 to help upgrade the state’s unemployment insurance system.
While the talent crunch is real, corporations are also considering outsourcing to access contractual mechanisms that ensure compliance, security and continuity of service requirements and to reduce the total cost of ownership by optimizing software licenses and lowering operating costs. In addition, outsourcing can enable parallel transformation activities that must run on the mainframe and new cloud infrastructure, address in-house skills gaps and drive greater business value with increased agility and speed to market, all managed by contractual service levels.
The Era of New Outsourcing Models
Trends in mainframe usage – whether in-house or outsourced – vary by the size of the enterprise and the size of installed millions of instructions per second (MIPS) on the mainframe platform.
Small users, with fewer than 5,000 MIPS, are in fact the only enterprises that are reducing mainframe workloads, with a decrease of approximately five percent per year in favor of shared data center models or phased rehosting strategies. Medium users, with between 5,000 and 20,000 MIPS, are increasing mainframe workloads by around five percent per year, typically keeping strategic applications on the mainframe and moving non-core applications into the cloud, leveraging cost optimizations and tactical modernizations to reduce annual license costs. Large users, with 20,000 to 75,000 MIPS, are increasing mainframe workloads by 10 percent per year and investing in mainframe platforms, adding DevOps and automation in parallel with cloud options, and following a strategic mainframe roadmap that keeps the mainframe at the center of the enterprise landscape. The largest users, with more than 75,000 MIPS, are increasing mainframe workloads by 15 percent per year, while optimizing, modernizing and engaging third parties to drive transformation.
While few major banks have seriously considered moving their mission-critical workloads off the mainframe en masse, managed services providers are offering OpEx-friendly, on-demand models that supplant the heavy capital typically required so organizations can realize their performance goals without having to compromise on their technology or assume responsibility for end-to-end maintenance with a shrinking resource pool.
Just as we see mainframe technology evolving, we are also seeing two commercial models gaining prevalence in the market: traditional outsourcing leading to Mainframe-as-a Service (MFaaS) and remote management.
In a traditional outsourcing model, buyers typically transition all mainframe hardware, tape and storage operations to a supplier’s data center, where the supplier runs the mainframe environment on a managed service basis, billing the client monthly for capacity and storage based on agreed resource units and service levels. This model can involve the “sale” of client assets (hardware, software, resources) to the supplier, generating proceeds that may be used to drive parallel transformation to cloud infrastructure. In this model, the supplier has more scope to optimize and deliver the agreed-upon outcomes and can evolve delivery into an MFaaS model by operating a multi-tenanted environment with applications migrated and a pay-as-you-go pricing structure in place to reduce maintenance fees.
Buyers of remote management services retain responsibility and ownership of the hardware and software, while the supplier delivers technical and operational services remotely. This model can be built quickly and provide faster access to skills, but enterprises may secure fewer cost savings since the supplier has less control. Enterprises can still drive mainframe optimization and transformation by contracting specifically (and sometimes separately) with the supplier and managing them via critical deliverables and/or service levels.
Rather than moving away from the mainframe, financial institutions are now embracing this grand old technology at the heart of their broader digital transformation strategy. Whether it is through a hybrid cloud/mainframe model, the Mainframe-as-a-Service model, or a DevOps/AIOps model, firms have woken up to the fact that a multi-pronged strategy is probably the best way to achieve their technology and business goals. The key is determining which strategic levers to pull in what combination—a decision that will depend on each financial institution’s current environment and business imperatives.
One thing is for sure: The mainframe is not dead yet.