The situation
A UK Global Financial Institution running several hundred production Red Hat Enterprise Linux servers was carrying approximately £1,000,000 a year in Red Hat support, with a compliance review in progress.
The instinct when an audit lands is to get compliant fast: buy the licences, close the gap, move on. That is the right instinct for an IT team. It is the wrong outcome for a finance one.
The harder problem sat underneath it. Like most enterprise Red Hat estates, the commercial arrangement was cross-wired — discounts on adjacent Red Hat products written to depend on maintaining full RHEL support. Model an exit and the saving cancels itself out. We call that structure the Renewal Lever, and it is why most Red Hat exits die in the business case rather than in the technology.
What we found
The £500,000 of annual overspend had two distinct causes, and they behave differently.
Part of it was licence hygiene: subscriptions still billing against servers that had been decommissioned, consolidated or virtualised years earlier, plus support tiers nobody had revisited since purchase. That portion is recoverable without changing vendor at all.
The rest was the price of the support itself. SUSE Multi-Linux Support prices materially below equivalent Red Hat support for the same coverage on the same estate — typically around 50% on a like-for-like basis. That delta is structural. It does not depend on finding unaudited servers; where ghost subscriptions exist, they sit on top of it.
The approach — Gated Decoupling
The sequence matters more than the destination, because the Lever only bites if support is cancelled while the estate still depends on Red Hat elsewhere.
| Stage | What happens |
|---|---|
| Dependency Register | Every Red Hat coupling in the estate is mapped — including subscription-gated content and management tooling that only surfaces once a migration is already underway. Ghost subscriptions and tier mismatch surface here. |
| Parity Gate | Where automation is in scope, it moves to open-source equivalents and the customer's own engineers sign off parity. Nothing is cancelled and no notice is served until that gate closes. |
| Patch-Source Switch | SUSE Multi-Linux Support takes over patching of the existing RHEL hosts where they sit. No rebuild, no OS migration, no retraining. |
The Parity Gate is the point. It is what turns an exit from a bet into a sequence — the customer's own team decides when the route out is proven, and nothing irreversible happens before they do.
The outcome
Approximately £1,000,000 a year became roughly £500,000, with operational continuity maintained across the estate. Same servers, same engineering team, same runbooks.
"The transition required zero server changes, zero retraining, and zero operational disruption. The saving was immediate, recurring, and material." UK Global Financial Institution — anonymised under NDA · Reference available on request
"The reason most organisations don't achieve this isn't technical. It's structural. When an audit lands, the instinct is to get compliant fast — buy the licences, close the gap, move on. That's the right instinct for an IT team. It's the wrong outcome for a finance one. The technology was never the root cause. The room was." Martin Roskelly — Co-Founder, Paringa
Why this pattern repeats in regulated enterprises
Financial services, telecoms and other regulated industries carry the same structural exposure: Red Hat contracts negotiated years ago, cross-wired discounts on adjacent products, and IT teams under pressure to resolve findings quickly rather than well.
The negotiating position erodes quickly once findings are formally issued. Audit and renewal cycles frequently overlap, and that is rarely a coincidence. The organisations that get a different outcome are not the ones that negotiate harder — they are the ones that arrive already holding a costed, sequenced route out.