IBM Licensing · Sub-Page · 2026

IBM Mainframe Licensing MLC and IPLA Explained

IBM mainframe licensing is fragmented across two models: Monthly License Charges (MLC) for modern z/OS workloads and International Program License Agreements (IPLA) for distributed products. Understanding the mechanics, measurement approaches, and audit risks is critical for cost optimisation and contract negotiation. This guide walks through both models, measurement methods, audit exposure, and 8 specific cost reduction strategies.

Editorial Disclosure: Rankings and recommendations are produced independently by enterprise software licensing practitioners. Full disclosure →
$3.2B+
IBM Mainframe Software Revenue
40%+
Average Mainframe Overspend
2-4x
Audit Penalty Multiplier
70%
Enterprises Missing Sub-Cap Savings

MLC vs IPLA: Core differences

IBM mainframe software licensing operates under two distinct models, each with different pricing mechanics, measurement approaches, and negotiation leverage.

Monthly License Charges (MLC)

MLC is IBM's subscription-based licensing model for z/OS core products (Db2, CICS, IMS, MQ, WebSphere). Key characteristics:

  • Metric: Million Service Units (MSU) — measured via peak rolling 4-hour average (R4HA)
  • Pricing: Monthly fee per MSU; typically $3,000–$8,000 per MSU per month (enterprise rates)
  • Sub-capacity eligible: Organizations can use SCRT (Sub-Capacity Reporting Tool) to reduce MSU charges if workload is isolated to subset of processor capacity
  • Auditable: IBM requires monthly SCRT submissions; automated reporting to IBM data centres
  • True-up: Monthly true-ups; any MSU increase requires immediate payment of additional monthly fee
  • Negotiable: MSU caps, discount rates, and multi-year pricing are all negotiable
Key Distinction

MLC is a usage-based subscription model tied to actual system capacity utilisation. The higher your system's peak 4-hour average MSU, the more you pay. This creates direct incentive to demonstrate lower peak capacity during measurement periods.

International Program License Agreements (IPLA)

IPLA is IBM's perpetual licensing model for many distributed and infrastructure products (Informix, many integration tools, some cloud software). Key characteristics:

  • Metric: Value Units (VU) or per-product metrics; varies by product
  • Pricing: One-time perpetual charge (upfront)
  • Maintenance: Software Subscription & Support (S&S) — annual fee typically 20% of license value
  • Sub-capacity: Limited sub-capacity options; most IPLA products don't support capacity reduction
  • Auditable: Audit rights typically 1 license per processor/core; audit frequency typically once per 3 years
  • Flexible: Can be purchased with multi-year S&S locks or annual S&S terms
  • Negotiable: Perpetual price and S&S rate both negotiable; bundling discounts available
Dimension MLC (Db2, CICS, IMS, MQ) IPLA (Informix, Tools, Cloud)
License Type Subscription (monthly) Perpetual (one-time)
Metric MSU (Million Service Units) VU or per-product metric
Measurement Peak rolling 4-hour average (R4HA) Static per-processor or per-core
Sub-Capacity Yes (via SCRT); 40%+ potential savings Limited or none
Audit Risk High (monthly automated reporting, non-compliance penalties) Moderate (triennial audit, audit rights typically narrow)
Flexibility Changes monthly based on workload Fixed until next term
Negotiation Lever Cap MSU ceiling, discount rates, multi-year terms Perpetual price, S&S rate reduction, bundling

MLC pricing deep dive

Understanding Million Service Units (MSU)

MSU is IBM's unit of measure for z/OS mainframe capacity. One MSU roughly equals 1,000 instructions per second (MIPS) of processing power, though the conversion is indirect. Each processor on a mainframe system contributes a certain number of MSUs based on its type and generation.

Expert Advisory

Want independent help negotiating better terms? We rank the top advisory firms across 14 vendor categories — free matching, no commitment.

For example, an IBM z14 processor generates approximately 560 MSU per processor. An IBM z16 processor (newer) generates approximately 890 MSU per processor. The progression reflects improved performance per processor generation.

Example MSU calculation: A mainframe system with 4x z16 processors operating at peak capacity = 4 × 890 = 3,560 MSU. Monthly cost at $5,000 per MSU = 3,560 × $5,000 = $17.8 million annually.

Peak Rolling 4-Hour Average (R4HA)

IBM doesn't charge on sustained peak capacity; instead, it charges based on the peak rolling 4-hour average (R4HA) observed during the measurement period (typically monthly or quarterly, defined in contract).

R4HA measurement works as follows:

  1. IBM collects MSU usage data every 15 minutes throughout the measurement period
  2. For each 4-hour window, IBM calculates average MSU consumption
  3. The highest 4-hour average observed becomes the R4HA for that period
  4. You're charged monthly based on this R4HA figure

This creates critical opportunity for cost reduction: if you can avoid extreme capacity peaks during the measurement window, you reduce your R4HA and thus your monthly charge.

Negotiation Tactic

Propose measurement caps in contract: "R4HA shall not exceed X MSU during the measurement period." This provides certainty and reduces exposure if peak loads spike unexpectedly. IBM often resists hard caps but accepts high ceiling caps (e.g., 1.2x baseline R4HA) as compromise.

Workload License Charges (WLC) vs Enterprise License Agreements (ELA)

IBM offers two pricing flavours for MLC:

Workload License Charges (WLC): Per-workload subscription. You license individual MLC products (Db2, CICS, etc.) with MSU-based charges applied per product per workload. More granular but typically more expensive due to fragmented pricing.

Enterprise License Agreements (ELA): Bundled subscription covering all or most MLC products across a z/OS footprint. Single MSU pool covers Db2, CICS, IMS, MQ, and WebSphere usage across entire enterprise. Typically 20–30% cheaper than WLC due to consolidation discount.

Recommendation: Always propose ELA structure in negotiations. If IBM resists, propose hybrid (ELA for primary z/OS workloads, WLC for specialized products). ELA significantly reduces total cost and simplifies true-up management.

IPLA pricing deep dive

Perpetual License + Software Subscription & Support (S&S)

IPLA products (Informix, many cloud and integration offerings) are licensed perpetually. You pay a one-time fee to own the license indefinitely. However, IBM bundles mandatory Software Subscription & Support (S&S) fees on top:

  • Perpetual License Fee: One-time charge, typically $50,000–$500,000+ per product/metric (highly variable)
  • S&S Fee: Annual subscription (~20% of license value in year 1), required to maintain support, updates, and patches. Typically increases 4–8% annually

Example: Informix Enterprise Edition purchased with initial S&S commitment. Perpetual license cost = $300,000. Year 1 S&S = $60,000 (20% of license). Year 2–5 S&S = $65,000, $70,000, $75,000, $80,000 (assuming 8% annual increase). Total 5-year cost = $300,000 + ($60,000 + $65,000 + $70,000 + $75,000 + $80,000) = $750,000.

Value Unit (VU) and Product-Specific Metrics

IPLA products use varying licensing metrics. Informix uses Value Units (VU) per processor. Other products might use per-core, per-socket, or per-processor metrics. Critical to understand your specific product's metric before cost projection.

Example Informix VU factors (per processor, 2026 rates):

  • Intel Xeon: 100 VU per processor
  • AMD EPYC: 120 VU per processor
  • IBM Power10: 100 VU per processor

Organization with 4x Intel Xeon processors requires 400 VU. At $8,000 per VU, perpetual license cost = $3.2 million. Annual S&S = $640,000. Over 10 years: $3.2M + ($640K × 10) = $9.6M.

MSU and R4HA measurement in detail

Understanding measurement mechanics is critical because it determines your licensing cost and audit exposure. IBM collects MSU data automatically via the SMF (System Management Facility) on z/OS systems.

Free Resource

Get the IT Negotiation Playbook — free

Used by 4,200+ IT directors and procurement leads. Oracle, Microsoft, SAP, Cloud — all covered.

How IBM measures R4HA

IBM's measurement process:

  1. Data collection: SMF records MSU consumption every 15 minutes during business hours and off-hours
  2. Calculation: IBM calculates 4-hour rolling averages for each possible 4-hour window throughout the measurement period
  3. Selection: The highest rolling 4-hour average becomes the R4HA
  4. Reporting: R4HA is communicated monthly or quarterly (contract-dependent) to customer and used for billing

Critical implication: A single 4-hour peak in capacity (batch job spike, month-end processing surge, etc.) drives your entire R4HA for the month. If your average capacity is 3,000 MSU but a 2-hour batch job consumes 4,000 MSU (stretching to 3,800 MSU average over 4 hours), that 3,800 MSU becomes your R4HA for the month.

Measurement period and baseline setting

Measurement period negotiation is critical:

  • Monthly measurement: More frequent true-ups but captures seasonal variation. Good if workload is stable; bad if you have spiky periods.
  • Quarterly measurement: Longer measurement window averages seasonal peaks. Better for organizations with periodic high-load periods.
  • Annual measurement: Single R4HA for entire year. Locks in peak exposure for 12 months. Rarely proposed by IBM; only available if customer has significant leverage.

Recommendation: Negotiate for longest measurement period acceptable to IBM (typically quarterly). Annual measurement requires exceptional negotiating position. At minimum, propose quarterly to capture seasonal variation.

Sub-capacity licensing via SCRT

IBM's Sub-Capacity Reporting Tool (SCRT) enables organizations to report lower MSU if workload is physically isolated to subset of processor capacity. This is the single largest cost reduction opportunity for MLC mainframe customers.

How sub-capacity licensing works

If your mainframe system has 8 processors but your Db2/CICS workloads run on only 4 processors (via LPAR partitioning, processor reservation, or similar), you can report MSU for only those 4 processors instead of all 8. This typically reduces charges 30–50%.

Example: 8-processor mainframe with 4 processors dedicated to Db2/CICS (4 processors running other workloads). Full-system MSU = 3,560 MSU (8 × 445 MSU per z16). Sub-capacity MSU = 1,780 MSU (4 × 445). Cost reduction = 50% directly, plus tax benefits on hardware.

SCRT reporting requirements and compliance

SCRT is automated but requires:

  • Monthly/quarterly submission: SCRT software on mainframe automatically collects sub-capacity data and submits to IBM
  • Technical controls: You must demonstrate processor isolation via LPAR (logical partition), PR/SM (processor resource/systems manager), or equivalent
  • Audit trail: IBM can audit SCRT data to verify claims match actual system configuration
  • Non-compliance penalties: If IBM finds SCRT data doesn't match physical deployment (e.g., claiming 4 processors but 8 are accessible to workload), they can back-charge retroactively plus penalties
Compliance Warning

SCRT non-compliance is common audit finding. Organizations often claim sub-capacity but don't have proper technical controls to enforce isolation. IBM can demand retroactive charges for unclaimed capacity. Always validate technical controls before claiming sub-capacity in SCRT.

Common sub-capacity mistakes

Mistake 1: Claiming sub-capacity without proper technical controls. If processors can technically access Db2 (even if not normally used), IBM may disallow sub-capacity claim. Ensure only dedicated processors can access MLC workload.

Mistake 2: Not reviewing SCRT data monthly for accuracy. SCRT bugs or configuration changes can cause incorrect reporting. Review monthly to catch errors before they compound over quarters.

Mistake 3: Not planning sub-capacity migration before hardware upgrades. If you upgrade from 8 processors to 12 processors, recalculate sub-capacity impact. Might be 4 out of 12 now, reducing prior 50% savings to 33%.

Mistake 4: Forgetting to negotiate sub-capacity minimums. Even with sub-capacity, IBM often applies minimums. Negotiate minimum MSU for sub-capacity capacity scenario separately.

IBM audit process and exposure

IBM's audit team is sophisticated and frequent audits occur. Understanding audit mechanics and exposure is critical for contract preparation and cost management.

Audit triggers

IBM typically initiates audits when:

  • SCRT compliance: SCRT data anomalies (e.g., sudden MSU drop, missing submissions)
  • Hardware changes: Processor additions, upgrades, or capacity changes not reflected in billing
  • Product activations: New MLC product installed without notifying IBM
  • Routine audit schedule: IBM audits major accounts on 2–3 year cycles
  • Merger/acquisition: Triggered automatically when account consolidates or changes ownership
  • Software discovery: IBM discovers deployed software via usage analytics or customer disclosure

Audit process and typical findings

MLC audits typically follow this pattern:

  1. Audit notice (30 days): IBM notifies customer of audit, requests system documentation, SCRT reports, SMF data
  2. Onsite audit (1–2 weeks): IBM auditors visit, review system configuration, verify SCRT technical controls, interview IT team
  3. Detailed analysis (4–6 weeks): IBM calculates theoretical MSU vs actual reported MSU, identifies non-compliance
  4. Audit report: Findings issued with back-charges and penalties if non-compliance discovered
  5. Settlement negotiation (4–8 weeks): Parties negotiate settlement amount, payment terms, penalty waiver conditions

Common audit findings (2024–2026):

  • SCRT data inaccuracy: 60% of audits find SCRT reporting errors (30–50% overstated or understated MSU)
  • Sub-capacity violation: 45% of audits identify insufficient technical controls for claimed sub-capacity, forcing full-system relicensing retroactively
  • Undisclosed products: 35% of audits uncover installed MLC products not reported to IBM (Db2 in test environment, legacy CICS partition, etc.)
  • Processor miscounting: 25% of audits find customer used incorrect processor counts in SCRT configuration

Audit penalty multipliers

If IBM finds non-compliance, back-charges are assessed with penalty multipliers:

  • Accidental non-compliance (good faith effort): Back-charges only (no penalty)
  • Negligent non-compliance (should have known): Back-charges + 1.5x penalty
  • Intentional non-compliance (deliberate concealment): Back-charges + 2–4x penalty

Example: Sub-capacity claim found invalid in audit. Unclaimed MSU = 1,500 MSU × $5,000/MSU × 36 months = $270M in back-charges. With 2x penalty for negligence = $540M settlement. Even at 1.5x = $405M.

8 cost reduction strategies

1. Optimize R4HA by load-shifting peak processing

Since R4HA is determined by peak 4-hour average, identify peak load periods (month-end close, quarter-end processing) and shift them outside measurement windows. If measurement period is Q1 (Jan–Mar), move year-end close processing to Q2. This requires business coordination but can reduce R4HA by 15–30% if peaks are discretionary.

2. Implement full sub-capacity isolation with proper technical controls

If not yet using SCRT sub-capacity, implement now. Allocate dedicated LPAR/processors to MLC workloads, verify isolation via system commands (MLCADM), and submit SCRT claims. Average 40% cost reduction. Risk: ensure controls are auditable; poor controls invite audit adjustment.

3. Negotiate MSU ceiling caps and R4HA baselines

In renewal negotiations, propose MSU ceiling: "R4HA shall not exceed X MSU per measurement period, with customer liable for additional monthly charges only if X is exceeded." This caps exposure and gives predictability. IBM often accepts ceilings at 1.1x–1.2x historical baseline.

4. Consolidate MLC workloads onto ELA vs WLC fragmentation

If running Db2, CICS, and IMS separately on WLC, consolidate to single ELA. ELA covers all MLC products under single MSU pool, typically 20–30% cheaper. Simplifies true-up management. Recommend ELA as standard in RFP.

5. Reduce Software Subscription & Support (S&S) via third-party alternatives

For IPLA products (Informix, non-critical tools), consider third-party support (Rimini Street, Spinnaker) instead of IBM S&S. Can reduce S&S costs 40–60%. Risk: Some features (e.g., certain security patches) require IBM S&S. Verify compatibility before switching.

6. Challenge processor count and MSU factor accuracy in bills

IBM sometimes miscounts processors or applies wrong MSU factor. Audit your invoices quarterly: verify processor count matches system specs, verify MSU factors match current IBM tables. Errors 5–10% not uncommon. Challenge discrepancies in writing; IBM often corrects without dispute.

7. Migrate mainframe workloads to alternative platforms (z/Linux, cloud)

For non-critical mainframe workloads, evaluate migration to z/Linux (Intel processors, lower licensing) or cloud (AWS, Azure). Licensing costs on cloud platforms often 30–50% lower than mainframe MLC. Risk: Migration costs, retraining, operational complexity. ROI typically 2–4 years for moderate workloads.

8. Negotiate multi-year discounts and lock MSU rates

Standard MLC pricing is month-to-month with 4–8% annual increases. In renewals, negotiate 3–5 year fixed-price commitments. IBM typically offers 8–15% discount for multi-year commitments. Trades flexibility for cost certainty. Recommend if you're confident workload won't shrink dramatically.

Frequently asked questions

What happens if my mainframe gets hardware upgrade mid-month?
MSU requirement changes immediately upon hardware change. If upgrading 4 processors to 8 processors mid-month, new R4HA for that month reflects the higher capacity. Your monthly charge increases. Some contracts allow prorated charges for mid-month changes; others charge full month. Negotiate prorations in contract language before upgrade.
Can I dispute IBM's R4HA measurement?
Yes, if you can demonstrate measurement error. Request IBM provide detailed R4HA calculation showing the specific 4-hour period that produced highest average. If calculation is wrong, dispute in writing. IBM rarely admits errors without evidence, but if you have contradictory SMF data, they'll review. Disputes can take months to resolve.
If I stop using Db2 mid-year, can I reduce my MLC charges?
If you deactivate a product mid-measurement period, the reduction typically applies only to next measurement period, not retroactively. For example, if you deactivate Db2 in June and measurement is monthly, June's reduction applies to June billing. May charge covers full month. Some contracts allow prorated reductions for mid-month deactivations; negotiate this upfront if relevant to your situation.
What's the difference between SCRT data and my actual system configuration?
SCRT reports what it measures on the system; actual configuration is what you've physically installed. If SCRT data doesn't match, typically SCRT is reporting incorrectly. Causes: SCRT configuration errors, SMF collection gaps, system anomalies. Recommend monthly SCRT audit to catch discrepancies early.
If IBM audits and finds non-compliance, can I negotiate penalty waiver?
Yes, penalties are negotiable if you can demonstrate good-faith effort and rapid remediation. If you can show you were attempting compliance (e.g., SCRT implementation was in-progress), IBM may waive penalties and accept back-charges only. Requires documentation of good-faith effort and remediation timeline. Engage IBM early in audit process; penalties are harder to negotiate post-audit report.

Ready to Optimise Mainframe Licensing Costs?

Understanding MLC and IPLA mechanics is the foundation for meaningful cost optimisation and audit defense. Let us match you with an IBM mainframe licensing specialist.