Last Tuesday, IBM finally unveiled its much-hyped z17 mainframe – and they had much to be excited about with this new mainframe. Like everything else these days, it comes with AI – or, more accurately, it comes with chips designed for AI. The z17s become generally available on the 18 June, with the IBM Spyre Accelerator to follow in the final quarter of 2025. Let’s take a look at the details of the new machine.
We first heard
about IBM Telum II chip and the Spyre Accelerator back at the start of
September last year, when it was unveiled at the Hot Chips 2024 conference.
They told us then that the chips are designed to support a broader, larger set of
models with what’s called ensemble AI method use cases. Using ensemble AI
leverages the strength of multiple AI models to improve overall performance and
accuracy of a prediction as compared to individual models.
The Telum II
comes with an on-chip AI accelerator, as you’d expect, so the z17 has the
ability to process 50 percent more AI inference operations per day than the
z16. In fact, it can perform over 450 billion inferencing operations daily with
one millisecond response times. There’s also been a 40 percent growth in cache,
which goes some way towards making this possible.
The IBM Spyre
Accelerator comes on a PCIe card, and provides additional AI compute
capabilities, creating environments to support multi-model methods of AI. The
Spyre Accelerator is specially engineered to bring generative AI capabilities
to the mainframe including running assistants, leveraging enterprise data
contained in the system.
Among the AI
assistants and AI agents that are supported are IBM watsonx Code Assistant for
Z and IBM watsonx Assistant for Z. These enhance developer and IT operations
productivity. For the first time, watsonx Assistant for Z will be integrated
with Z Operations Unite to provide AI chat-based incident detection and
resolution using live systems data. IBM Z Operations Unite is a solution
designed to unify key performance metrics and logs using OpenTelemetry format.
This is expected to accelerate the time to detect anomalies, isolate the impact
of potential incidents, and reduce the resolution time.
Ross Mauri,
general manager of IBM Z and LinuxONE, IBM said: “With z17, we're bringing AI
to the core of the enterprise with the software, processing power, and storage
to make AI operational quickly. Additionally, organizations can put their vast,
untapped stores of enterprise data to work with AI in a secured, cost-effective
way.”
In, what might
be called, a teaser trailer, IBM previewed a new version of its z/OS operating
system (V 3.2), which will be released in the third quarter of 2025. z/OS 3.2 will
support hardware-accelerated AI capabilities across the system and operational
AI insights for system management capabilities. What’s more, z/OS 3.2 will
support modern data access methods, NoSQL databases, and hybrid cloud data
processing, helping AI software tap into a broader set of enterprise data and
derive predictive business insights.
Of course, AI
isn’t the only thing on the minds of potential customers of the z17. The big
issue for many mainframe-using organizations is security. IBM has addressed
that worry in the z17 announcement.
The z17 also
includes IBM Vault, which will help to standardize secrets management across
hybrid cloud, leveraging technology from HashiCorp, which were announced last
month. IBM Vault uses identity-based security to authenticate and authorize
access to secrets, certificates, keys, tokens, and other sensitive data. With
the addition of IBM Vault, clients can have a single solution to help protect
critical workloads by managing the entire secrets lifecycle across their full
IT estate.
IBM also intends
to deliver new capabilities for discovering and classifying sensitive data on
the platform. This would tap into Telum II and utilize natural language
processing so that mission-critical data can be identified and protected. In addition,
IBM Threat Detection for z/OS, a new AI-driven security solution, is designed
to detect and identify potentially malicious anomalies that might be the result
of a cyber-attack.
IBM Support for
z17 helps clients optimize their environments for peak performance to address
risk and disruptions for mission-critical operations. IBM's AI processes
streamline incident remediation and help improve case resolution time, built on
IBM watsonx, now support IBM Z systems.
IBM says that
the tenth generation of IBM Storage DS8000 is designed to harness the full
power of IBM z17, providing organizations access to critical workloads,
consistent and optimized data performance, and a modular architecture to adopt
the latest IBM research-backed technologies to fuel business growth while
monetizing data.
The z16 has
been selling well for IBM over the past couple of years, and investors seem to
anticipate much the same for the z17, with a 2% rise in IBM's stock following
the announcement.
It certainly
seems like IBM is moving in the right direction with this latest mainframe
announcement.
