① Guide · industry · airlines and travel
Airlines, GDS operators, and travel platforms run z/TPF and z/OS at some of the highest transaction rates in computing. The usual technical levers, soft capping and workload shifting, are off the table when throttling risks the booking flow. So the cost lever moves to the contract. Here is where it sits.
Extreme volume, non-negotiable workload, spiky demand. A hard licensing profile.
The reservation and operations core of many carriers and global distribution systems runs on z/TPF, IBM's Transaction Processing Facility, a lineage that traces back to the original SABRE project and was engineered for extreme volume, very high concurrency, and sub-second response. Around it sit z/OS workloads, Db2 for z/OS, and tooling from Broadcom (CA), BMC, and other publishers. The defining trait for licensing is intensity: these are among the busiest systems in existence, so capacity based charges concentrate quickly and every MSU is doing mission critical work.
That intensity removes levers most estates rely on. A departure control or booking system cannot tolerate the soft capping or workload shifting that trims MSU peaks elsewhere, because any throttling threatens the transaction flow itself. Travel demand is also sharply seasonal and event driven, so a single peak month, a holiday rush or a disruption recovery, can set a baseline the rest of the year never justifies. With the technical knobs largely locked, the cost discipline shifts onto contract structure, where it can still be considerable.
| Lever | Why it is constrained here | The buyer side move |
|---|---|---|
| Soft capping | Throttling risks the booking flow. | Price the peak instead of capping it. |
| Workload shifting | Core TPF work cannot move off platform. | Right size tiers to sustained load. |
| Seasonal peaks | Holidays and disruptions spike demand. | Surge terms so a peak month is not the baseline. |
| Pricing model | High, stable base load is the norm. | Choose the model that suits sustained volume. |
| Capacity validation | Billed peaks drive a very large bill. | Validate the peak independently before renewal. |
Patterns commonly observed for high intensity travel workloads. The specific products, metrics, and peak handling that apply live in each agreement, which is why the work starts from your own contracts and consumption data.
The core TPF and reservation workload is treated as fixed, and effort is directed where it can actually move: the contract terms, the capacity tiers, and the peak handling. No proposal risks the booking flow to chase a saving.
Protect the transaction, negotiate the contract.
Entitlements are sized to genuine sustained demand, with seasonal and event surges handled as exceptions in the terms rather than baked into a permanent baseline. A single busy month should not set the price for twelve.
The base load pays; the spike should not.
On systems this large, a small error in the billed peak is a large number. The capacity that drives the charge is validated independently, LPAR by LPAR, before any renewal is priced off it.
At this scale, small errors are big money.
Caps, surge provisions, and protections for disaster recovery and seasonal capacity go into the agreement, so the contract handles the realities of travel demand rather than punishing them.
The contract should expect the spike.
④ What changes with us in the room
You cannot throttle a booking to save money. So we move the saving into the contract.
Typical renewal reduction
Engagements delivered since 2019
Mainframe spend negotiated on the buyer side
The reservation and operations core of many carriers and GDS operators runs on z/TPF, IBM's Transaction Processing Facility, built for extreme volume and sub-second response. Around it sit z/OS workloads, Db2 for z/OS, and tools from Broadcom (CA), BMC, and others. The defining trait for licensing is intensity: very high transaction rates concentrate capacity charges fast.
Because the workload is enormous and non-negotiable. A reservation or departure control system cannot tolerate the soft capping that lowers peaks elsewhere, since throttling risks the booking flow. That removes technical levers and shifts the emphasis onto contract structure, and demand is spiky, so a peak month can lock in cost the rest of the year never justifies.
By moving the work from the technical layer to the contractual one. Where you cannot shape the peak safely, you negotiate how it is priced: tiers sized to sustained demand, a model that suits a stable base load, and protections against seasonal spikes setting a permanent baseline. Right size to real consumption, validate the billed capacity, and build surge handling into the term.
It belongs in the negotiation, not as an afterthought. Travel operations carry stringent continuity requirements, and DR capacity has its own licensing treatment that varies by site posture. The terms governing hot, warm, and cold standby should be set deliberately so recovery capacity is covered without paying full production rates for capacity that rarely runs.
Related: Mainframe licensing in telecom · Mainframe licensing in healthcare · DR licensing rules · IBM licensing hub · cost optimization service
Audit notice or renewal under 18 months out? We mobilize within 48 hours.
Get expert help →