In 2024, the mainframe turned 60. Mainframes are used by 71% of the world’s Fortune 500 companies, according to IBM, and handle 90% of credit card transactions. They are used by 44 of the top 50 banks globally, carry 68% of the world’s IT production workloads, and are accountable for 6% of IT expenditure.
Forrester has said that the mainframe is only going to gain momentum over the next two years; its ‘State of Mainframes Global, 2024’ report said that 54% of respondents were going to expand their mainframe use cases. Only 15% were planning on reducing their reliance on this technology.
According to the BMC’s ‘State of the Mainframe in 2024’ survey, 94% of respondents believe it to be a long-term platform capable of handling new workloads – a statistic consistent with perceptions of 93% since as far back as 2003.
The system is also scalable, secure, aligned with growing compliance expectations, and provides capabilities for both legacy and emergent technologies. Furthermore, it’s reliable – the availability of the mainframe can exceed 99.999%.
“The mainframe has unmatched reliability, security and performance – it is designed to operate 24/7, can detect and recover from hardware failures, minimising downtime, and has significant security features,” says Jochem Melis, markets CIO, Absa. “It has also consistently evolved, enhancing resilience, expanding processing power, and reducing its form factor. The first mainframe weighed about five tonnes and occupied a room, while the latest [IBM] Z16 models take up a single rack in the datacentre.”
Brent Ellis, senior analyst, Forrester, says that many companies, particularly financial institutions, built a lot of their core functionality on top of their mainframes, upgrading through the years. “These upgrades are to the point where a substantial amount of compute processing required to close transactions in the time everyone needs is so big, that the mainframe has evolved into one of the very few platforms where this can be done at scale.”
Intelligent solutions
Part of it is the technology. The other part, says Ellis, is the range of comprehensive services provided by IBM to customers around the technology. “It’s built an entrenched moat around the technology,” he says. “Platforms on the mainframe are incredibly risk-sensitive. They require reliability, and as far as technology platforms go, mainframes still represent one of the most reliable you can buy. If you look at a lot of instances in AWS or Azure or another hyperscaler, you’re looking at creating architecture patterns that basically emulate what you’re getting in the mainframe.”
And mainframes continue to adapt. The IBM Z series can handle AIOps, blending AI and machine learning algorithms to automate processes and manage complex IT environments.
Jaco Sadie, mainframe modernisation leader for Middle East and Africa, IBM, says: “With the excitement surrounding generative AI, you can assume legacy IT infrastructure like the mainframe has not been invited to the party, but our IBM Institute for Business Value study found that 79% of IT executives believe the mainframe is essential for enabling AI.” A large percentage of AI deployments require that organisations move their data to the cloud, but for industries that rely on high-speed data processing to handle highly sensitive data, keeping AI capabilities closer to where the data resides offers plenty of advantages – and the 84% of respondents to the IBM survey agree. Infusing AI into transactional workloads to get real-time insights into business applications can deliver faster decision-making and give companies the freedom to create intelligent solutions that enjoy the benefits of AI but protect the data.
“AI is influencing how we use the mainframe,” says Jörg Fischer, CIO, Standard Bank. “While the mainframe is primarily a transaction processing system, AI can be integrated for realtime processing. We can perform real-time checks using AI on the mainframe, using data lakes and data stores for enhanced capabilities.”
Combining AI’s ability to identify issues with the mainframe’s ability to detect operational anomalies is a useful business case; using AI for mainframe operating systems and subsystems could predict resource usage and outages while extracting insights about system performance. “Around 78% of IT executives are piloting projects or operationalising initiatives that include AI into mainframe applications and transactions,” says Sadie.
Absa, meanwhile, is using hybrid cloud models to extend the capabilities of the mainframe onto cloud platforms that are scalable for customer-facing applications, while the mainframe handles high-volume transaction processing. Absa’s Melis says it has an API-driven architecture that decouples core functions with modern, cloud-native services.
Our core banking system still runs on the mainframe – we are not migrating these systems to the cloud; we are transforming them to be cloud-ready.
Jörg Fischer, Standard Bank
Standard Bank’s Fischer says its mainframes have been a reliable and scalable system for the bank for many years. “While we are modernising our systems and moving to the cloud, we’re not migrating off the mainframe for the sake of it. Instead, we are transforming our infrastructure to integrate with cloud technologies. For example, the next version of our SAP core banking system will be cloud-ready and we’re working on a multiyear project to transition to this version in the cloud. It is one of the most critical technology assets that reside in our datacentres.”
Steady transition
IBM thinks a hybrid approach offers the most business value, and calls this architectural framework “hybrid-by-design”.
“The requirements of the business can drive technology decisions instead of technology issues forcing the strategy,” says IBM’s Sadie. “Companies that have embraced the hybrid-by-design strategy, according to our study, report three times higher return on investment.”
Standard Bank’s mainframes have been running for more than 30 years, and it continues to use them to manage a large number of transactions and processes. “Our core banking system still runs on the mainframe – we are not migrating these systems to the cloud; we are transforming them to be cloud-ready. This allows us to use modern technologies while staying reliable and scalable,” says Fischer.
Absa, meanwhile, has made significant investments over the years in highly customised, in-house applications designed to benefit from the mainframe’s millisecond processing speeds. This makes a like-for-like replacement not a straightforward alternative.
“Many applications have been fine-tuned over decades, optimised for performance and security that is difficult to replicate elsewhere,” says Melis.
Forrester’s Ellis says he’s talked to many companies, specifically financial organisations, that are looking to modernise their mainframe workloads. “When they start to dig into the amount of time and effort required to migrate, they’re looking at five years, possibly more.”
The average tenure of a CIO is around three years, so if modernisation programmes extend beyond this, it’s hard to bring projects to completion. Even before AI erupted onto the scene, the conversation around the mainframe was changing – largely due to the realisation that the solution to modernisation wasn’t moving everything off the mainframe. Hybrid gives companies the freedom to explore cloud and AI while benefitting from the mainframe’s transactional workload capabilities.
“The mainframe has longevity within IBM,” says Ellis. “It has a strategy to enable more modern workloads to run on the hardware and the number of people acquiring mainframe hardware to run on Linux is increasing. Over the next few years, it is likely there will be more capability coming in to first Red Hat, as IBM owns it, and other Linux distributions to ensure a steady and non-disruptive transition to modern environments within the mainframe.”
WHAT WOULD ABSA DO?
Absa has found the mainframe to be well-suited for consolidating and processing large workloads as it houses the end-to-end workflows in the same infrastructure. It also reduces external complexities, dependencies and latencies while optimising resource usage such as CPU, memory and I/O. This has improved performance, reduced idle processing for peak event processing and helped reduce energy consumption. Absa’s Jochem Melis says the mainframe has been integral to the transformation of the bank’s digital platforms.
“The mainframe has been pivotal in helping the organisation adapt to changes within the payments industry over the last few years, evolving it by moving from the historical batch-driven debit order collection processes to real-time DebiCheck collections processing, and adapting to the local peer-to-peer payments industry from real-time clearing to PayShap,” he says. “We can maintain our current interfaces and evolve to meet the changes of the market and technology while still providing the increase in transactional volumes brought on by increased digital adoption.”
* Article first published on brainstorm.itweb.co.za
Share