IBM updated its z/OS operating system, adding AI optimization and quantum-focused features to its z17 mainframe.

Update z/OS 3.2 brings features originally previewed by IBM in April, including AI-driven traceable network packet batching intelligence, and an AI assistant for optimizing workload management.

AI use in batching packets can better reduce CPU overhead by monitoring workload and traffic patterns, and adjusting batch size or timing to balance CPU usage and latency.

Ways to balance CPU usage across network and compute-intensive AI workloads can also be suggested by the Workload Manager (WLM) Policy Advisor, an AI assistant available in the z/OS update.

On the security front, the new OS comes with data access-based threat detection which uses AI to analyze behavioral patterns and identify potential threats in real time. Included in IBM Threat Detection for z/OS is a quarantine feature leveraging built-in security manager RACF to contain threats.

Network security could also see augmentation from various quantum-based features available to the mainframe, implementing National Institute of Standards and Technology (NIST) quantum-safe algorithms. While the exact algorithms aren’t named, probable contenders include the preferred NIST PQC standard HQC (Hamming Quasi-Cyclic).

HQC and other quantum-safe algorithms were included this week in an update to security firm Fortinet’s operating system, better preparing network managers for the future threat of a quantum computer becoming powerful enough to break widely used encryption methods.

Cloud and more

IBM has also added modern APIs to z/OS, which it says are engineered to make data within the OS accessible in hybrid cloud and AI environments without resorting to “costly and error-prone” Extract-Transform-Load (ETL).

IBM’s mainframe can also run Docker containers, and with built-in networking from the Communications Server Sysplex Distributor, traffic can be automatically routed to them without the need for an external load balancer. This potentially gives mainframe-level reliability, security, and performance for network traffic to cloud-style apps running on z17.

The z17, first unveiled in April and made available in June of this year, is touted by IBM as an AI-ready mainframe for today’s data processing needs.

It is powered by a second-generation on-chip AI accelerator built on the IBM Telum II processor, the cache capabilities of which enable a one-millisecond response time and more than 450 billion inferencing operations in a day.