Leveraging IBM Specialty Processors (zIIP, zAAP)
IBM mainframe Monthly License Charges (MLC) often rank among the largest ongoing IT expenses for enterprises. For organizations running IBM Z, controlling these costs is a top priority.
Specialty engines, such as zIIPs and zAAPs, offer a powerful way to reduce software spend by offloading specific workloads from general processors.
By moving eligible work to these dedicated engines, companies can reduce the chargeable MSUs that contribute to higher monthly IBM software bills. Read our ultimate guide to IBM Mainframe Licensing (z Systems): Models, Costs, and Negotiation Tips.
In this guide, we’ll explain how IBM specialty engines work, which workloads qualify for offloading, and how to maximize cost savings.
You’ll also learn about the return on investment (ROI) of deploying specialty processors, the potential risks to consider, and negotiation tips to ensure the savings stay in your pocket.
1. What Are IBM Specialty Engines?
IBM mainframes support specialty processors – dedicated engines for certain types of work. These processors run alongside general-purpose CPUs (CPs) but execute eligible workloads in a way that does not count toward your software usage charges.
The primary specialty engines relevant to IBM Z licensing are:
- zIIP (Integrated Information Processor): Offloads eligible z/OS workloads, including parts of DB2 (analytics queries, certain utilities, and DRDA distributed processing), Java and XML processing, and select network encryption or compression tasks. zIIPs free up general CPU capacity by handling these tasks separately, thereby lowering software-measured usage.
- zAAP (z Application Assist Processor): Originally designed to offload Java and XML workloads on z/OS. In modern systems, zAAP functionality has been merged into zIIPs. Essentially, zIIP is now the primary specialty engine that handles what zAAP used to cover.
- Other specialty engines: IBM also offers the IFL (Integrated Facility for Linux) for Linux-only workloads on Z, and the ICF (Internal Coupling Facility) for Sysplex coupling tasks. Like zIIPs, these specialty engines do not count toward z/OS MLC, though they serve specific non-z/OS purposes.
Key point:
Workloads running on specialty processors are excluded from IBM’s MLC MSU calculations.
CPU cycles on zIIPs, zAAPs, IFLs, or ICFs do not contribute to the peak MSU numbers that determine your software bill. This is why specialty engines are so valuable for controlling mainframe costs.
2. Cost Benefits of Specialty Engines
Deploying specialty engines can yield significant cost savings by lowering your billable mainframe usage. IBM calculates software charges based on peak CPU consumption (MSUs) on general processors.
Every unit of work you shift to a zIIP (or other specialty engine) is work not done on a general CP – directly reducing the MSUs counted for billing.
- Lower MLC bills: Offloading eligible workloads reduces monthly license charges. If you redirect even 10–20% of CPU activity to zIIPs (common when offloading DB2 analytics or Java), you can reduce your software bill by a substantial margin. Shaving down the peak MSU level means you avoid the costliest usage tiers.
- Rapid ROI: Specialty engines require an up-front hardware investment, but payback is often fast. Many organizations recoup the cost in 6–18 months through MLC savings. After that, the ongoing reductions in monthly fees are pure savings year over year.
Example: Consider a mainframe environment with a peak load of 1,000 MSUs (around $12 million per year in MLC fees). If you add zIIP processors and offload 20% of that work (200 MSUs), the peak on general CPs drops to 800 MSUs (about $9.6 million per year). That’s a reduction of roughly $2.4 million annually in software cost, as shown below:
| Scenario | Peak MSUs | Annual MLC Cost (approx.) |
|---|---|---|
| Before zIIPs (0% offload) | 1000 | $12 million |
| After 20% offload to zIIPs | 800 | $9.6 million |
| Estimated Annual Savings | – | $2.4 million |
Even after accounting for the one-time hardware expense of the zIIPs (say $2 million), the payback period would be under a year. Every year after that, the full $2.4 million in savings is allocated directly to reducing your IT costs.
To maximize the benefits, do your homework before investing in zIIPs:
- Workload analysis: Identify which applications and tasks are zIIP-eligible and measure their current CPU usage.
- Projected MSU reduction: Estimate the drop in peak MSUs after offloading those tasks, and translate that into monthly MLC savings.
- ROI Projection: Prepare an ROI analysis that compares hardware costs to expected savings over time (e.g., anticipated payback period).
Having a solid analysis and business case ensures you focus on the most impactful workloads and sets clear expectations for cost savings.
Use our checklist for Mainframe negotiations, IBM Mainframe Licensing Negotiation: Checklist for Success.
3. Eligible Workloads
Only specific types of work can be offloaded to specialty processors. Understanding which workloads are eligible is critical to planning your offload strategy.
Key categories include:
- DB2 for z/OS (analytics & utilities): Portions of DB2 can run on zIIP – for example, parallel query threads, certain index build or utility jobs, and servicing remote (DRDA) clients. Intensive analytical queries or batch DB2 jobs are prime candidates to move off general CPs and onto zIIPs.
- Java on z/OS: Java applications running under z/OS (such as WebSphere Application Server, Java batch jobs, and Java stored procedures in DB2) are eligible to execute on zIIPs. By shifting Java workloads to zIIPs, you dramatically reduce the MLC costs associated with those applications while still leveraging the mainframe’s performance and reliability.
- XML and web services: XML parsing and other XML/JSON processing in z/OS can be directed to zIIPs. For instance, z/OS XML System Services will use zIIP for parsing tasks if available. This means modern SOAP/REST web services or any XML-heavy processing on the mainframe can be handled more cost-efficiently.
- Encryption and compression: Certain cryptographic and data compression tasks can exploit zIIPs. A common example is IPsec network encryption on z/OS – some of that encryption workload can be processed on zIIP instead of a general CP. Similarly, some software-based compression routines (if not using a hardware accelerator) may be eligible for zIIP processing. Offloading these tasks keeps security and compression overhead from inflating your MSU usage.
- Middleware (e.g., IBM MQ): IBM MQ on z/OS can offload parts of its message processing to zIIPs. Other IBM middleware and tools have also gradually added zIIP support. Check each product’s documentation; enabling zIIP usage for messaging, integration, or utility software can immediately contribute to cost savings.
To get the most out of your specialty engines, ensure your system is configured to use them. Enable DB2 and other subsystems to take advantage of zIIP offload (e.g. turn on DB2 query parallelism, configure JVMs to use zIIP).
The goal is to push as much eligible work as possible onto zIIPs, leaving only ineligible tasks on the expensive general processors.
Insights on licensing models, IBM Sub-Capacity Licensing on Mainframe vs Distributed: What’s the Difference?.
4. Considerations & Risks
Before diving into a specialty engine strategy, consider these factors and potential risks:
- Hardware and capacity costs: You’ll need to purchase or upgrade your system to add zIIP processors. They are cheaper than general CPs but not free, so budget for that capital expense. Also, ensure your mainframe model can accommodate the number of zIIPs you plan to use, and size those engines to handle the volume of work you intend to offload.
- Workload eligibility limits: Not everything can run on a zIIP. Core online transaction processing (CICS, most IMS, traditional COBOL batch) will still run on general CPUs. This means there’s a ceiling on how much you can reduce MSUs – for example, if only 15% of your workload is offload-eligible, that’s the maximum reduction in CPU usage you’ll achieve. Specialty engines won’t eliminate MLC costs, but they can significantly reduce them.
- IBM oversight: IBM monitors your usage via the monthly SCRT report. If your reported MSUs drop sharply due to zIIP offloading, IBM might take notice. They could inquire or review to ensure you’re using zIIPs within allowed guidelines. To stay compliant, use zIIPs only for officially eligible work. It’s wise to document which workloads you moved to zIIP and why they qualify, so you can readily answer any questions.
- Performance and Spillover: Monitor performance after offloading. If your zIIPs are undersized or overburdened at peak, z/OS will shift overflow work back to general CPs (which means unexpected MLC charges and potential slowdowns). Prevent this by monitoring zIIP utilization and adding capacity if needed. The idea is to save money without compromising service levels – critical jobs should run smoothly on zIIP and not be redirected to a regular processor due to a lack of zIIP capacity.
5. Negotiation & IBM’s View
IBM’s stance on specialty processors can seem paradoxical. They encourage customers to use zIIPs (since IBM sells the hardware and it keeps workloads on the mainframe), but also know that offloading reduces their software revenue. The upshot: you need to be savvy in negotiations to capture the full benefits of zIIPs.
Keep these tips in mind when discussing zIIPs with IBM or revisiting your contracts:
- Confirm eligibility in writing: Obtain confirmation from IBM (via email or contract language) that your targeted workloads are zIIP-eligible. This avoids disputes later on and demonstrates to IBM that you’re informed about the rules.
- Seek upgrade credits: When purchasing zIIP hardware, ask if IBM offers any incentives or Technology Upgrade Credits. They might offer credits or discounts to support your investment in modernizing the system, effectively reducing your costs.
- Separate deals for hardware and software: Don’t let one hand give back what the other hand saves. In other words, avoid tying hardware purchases to software licensing deals in a way that dilutes your savings. Negotiate your hardware acquisition (zIIPs) independently for the best price, and separately ensure that your software pricing aligns with the reduced MSU usage. The goal is to see a real reduction in your invoice, not just a shifting of costs.
IBM ultimately wants to maintain your business on IBM Z, so use that fact.
By demonstrating that you can and will optimize costs (using IBM’s own offered tools like zIIP), you gain leverage. Be prepared, and be firm in expecting the MLC reductions; you can then negotiate from a position of strength.
6. FAQs
Q: How much MLC savings can zIIPs really provide?
It depends on your workload mix. Many shops offload 10–20% of CPU activity to zIIPs, which can translate into millions of dollars in annual MLC savings for a large environment. Most companies see full ROI within a year or two of implementation.
Q: Do I need to notify IBM if I move workloads to zIIP?
No formal notification is required when you start using zIIPs. IBM will see the effect in your monthly usage reports. If your MSU consumption drops significantly, they may ask about it during contract reviews. As long as the work moved is legitimately zIIP-eligible, you comply.
Q: Can all DB2 workloads run on zIIPs?
No. Only certain parts of DB2 are zIIP-eligible (for example, parallel queries, some utilities, remote DRDA requests). Regular transactions and most SQL queries still run on general-purpose CPUs, so core DB2 processing will always remain on standard processors.
Q: Is zAAP still relevant, or is everything on zIIP now?
zAAP engines are largely a thing of the past. IBM rolled zAAP capabilities into zIIP on modern systems. Today, a zIIP can handle what a zAAP used to (Java and XML workloads), so most organizations have consolidated on zIIPs. If you still have older zAAPs, IBM provides ways to convert or upgrade them to zIIPs.
Q: Are specialty engines subject to IBM audit or review?
IBM doesn’t audit zIIPs separately, but they do review overall usage. If your offloading is extremely high or unusual, IBM might investigate to ensure compliance. To prepare, use zIIPs as intended (for eligible work only) and keep records of your configurations. That way, if questions arise, you can demonstrate that you’ve adhered to IBM’s guidelines while optimizing your costs.
Read about our IBM Licensing Advisory Services
