Mainframes have always played a significant role in IBM’s business, and the company is now making strides to modernize its offerings. With the introduction of the z16, IBM is ushering its mainframes into a new era driven by AI and the advent of quantum technologies.
Mainframes Continue to Thrive
IT directors are often questioned about the relevance of mainframes, which are increasingly seen as technical debt. However, the real source of this technical debt lies in the COBOL and PL/I applications that continue to run on these machines, with more retired professionals having the necessary skills compared to newly graduated engineers. According to IBM, its zSystems still process 70% of global transactions by value. As a result, 45 of the top 50 banks, 8 of the 10 largest insurance companies, and 8 of the 10 largest Telcos continue to rely on mainframes. IBM claims their clients have shown no clear signs of wanting to part ways with their mainframes. When a mainframe process handles billions of euros on a daily basis, companies are understandably hesitant to recreate such a system. In fact, IBM cites a study stating that “90% of IT executives (with mainframes) consider their mainframe as a platform for growth,” with more than half reporting an increase in transaction volumes over the past 12 months. These figures clearly demonstrate that mainframes are still very much alive and actively used in data centers.
Increased Competition in the Cloud Era
With the rise of cloud computing, IT executives are understandably concerned about the future of mainframes. Furthermore, hyperscalers like AWS and Azure have their sights set on this potentially lucrative market. These cloud providers are developing reference architectures and services to help businesses migrate their mainframe code and applications to the cloud. The idea of having mainframes in the cloud is gaining traction, even within IBM itself. The company recently announced an offering called “IBM Z as a Service,” known as Wazi. However, this service primarily focuses on testing and development, aiming to empower mainframe application developers and not to replace the golden goose. This is evident in IBM’s legal battle with LzLabs, a Swiss company that has developed container technology to emulate the zOS environment on x86 servers and directly execute Z System binaries. LzLabs also offers a technology that converts COBOL code into Java, enabling enterprises to modernize their applications while still maintaining access to their source code. IBM claims that this violates its intellectual property and seeks to prevent the sale of these technologies, which could potentially lead to a wave of mainframe abandonments.
A Promising Future for Mainframes
While IBM may be actively combatting competition, its primary focus is on securing the future of its mainframes by modernizing their design and usage. The recently announced z16 range introduces a three-pronged strategy centered around security, AI, and DevOps. This new line represents a significant shift in hardware architecture, featuring a new processor and tools aimed at addressing these three key areas.
Security in the Quantum Era
According to Ric Lewis, SVP of IBM Systems, “IBM Z sets the standard for highly secure transaction processing.” To maintain this reputation, IBM is staying ahead of the curve. Recognizing that states and cybercriminals are already archiving stolen encrypted data, ready to crack it once quantum computers are powerful enough to execute Shor’s algorithm, IBM has integrated “post-quantum” encryption techniques into the z16. These techniques are designed to withstand the computational power of quantum machines. IBM explains that the z16 is protected by “quantum-safe” technology throughout multiple firmware layers during the boot process. It is the first system to support quantum-safe secure boot. Additionally, the Crypto Express8S adapter provides “quantum-safe” APIs, allowing companies to switch encryption technologies and modernize their applications for better data protection both in transit and at rest.
Another important consideration driving the development of the new mainframes is the need to integrate them into AI-driven analytics processes. One particular area of interest for IBM’s clients is fraud detection using AI. To fulfill this need, the z16 differs from its predecessors (z14 and z15) by adopting a redesigned processor architecture. The Telum processor incorporates a Neural Processing Unit (NPU) to accelerate inference tasks. With a computing power of 6 TFLOPS, IBM claims that the NPU improves response times by a factor of 20 and increases bandwidth by a factor of 19 compared to similar x86 servers. The initial applications for this NPU include real-time credit card fraud detection. The IBM z16 can analyze 300 billion inference requests per day with a latency of one millisecond.
Integration into the DevOps World
In its most robust configuration, the z16 features 32 Telum CPUs, totaling 256 compute cores operating at 5.2 GHz. Each of the four drawers in a complete z16 machine houses four sockets. This design mirrors the recent initiatives by Apple with its M1 Ultra and Nvidia with its Grace Superchip, where each socket incorporates a high-performance bus connecting two Telum CPUs. In other words, each socket hosts 16 compute cores (8 per Telum CPU) and 2 NPUs. This architecture delivers exceptional efficiency, with IBM claiming a 40% performance increase per socket compared to the z15. IBM now believes that such power should also benefit modern applications. The z16 can host Kubernetes clusters and supports Red Hat OpenShift for this purpose. This transformation turns the “transaction-devouring monster” into a “container ogre.” In theory, a complete z16 machine (with all four drawers filled) can simultaneously manage 3.5 million Docker micro-containers running microservices under NGinx. With this approach, IBM is pursuing a dual vision to ensure the future of its mainframes in modern development processes and DevOps chains. On one hand, the “Cloud Modernization Stack” aids enterprises in modernizing their existing mainframe systems by providing a platform for converting COBOL and PL/I code into Java. On the other hand, the z16 supports numerous open-source technologies used in the DevOps world, enabling seamless integration into CI/CD chains for container deployments. Supported technologies include Ansible, Python, Node.js, GoLang, and Anaconda. Ross Mauri, General Manager of IBM Z, explains, “Today, we are making data on the mainframe more accessible. With the arrival of OpenShift on the platform, we can now run containerized workloads and microservices on our mainframes. We embrace open-source technologies on the IBM Z platform and offer a unified development experience across the hybrid cloud. The increased capacity and power of the z16 will better handle modern workloads.”
This modernization effort may not drastically change how companies view mainframes, but it will undoubtedly ensure their longevity among current clients. IBM has also adopted a subscription-based and pay-per-use commercial model, aligning with the cloud philosophy.