Rocky Linux migration economics, costed honestly.
Rocky Linux migration economics is the cost model that turns a stated intention to leave RHEL into a buyer-side instrument. The model is useful even when the migration never executes; it is what makes the exit posture readable to the Red Hat account team. This note sets out what the model must price and where most internal estimates understate the bill.
What the model actually has to price.
Rocky Linux migration economics begins with a list of line items the typical internal exit plan does not contain. The plan in the drawer prices the obvious. It multiplies the RHEL subscription saved per system by the number of systems in scope and calls the result the saving. That number is the headline. It is not the migration economics. The migration economics is what remains after every other line in the bill has been entered, and the bill has more lines than most plans count.
The buyer-side reading is that the cost model is a document with two audiences. The first audience is the internal sponsor who has to commit budget and engineering capacity to the migration. The second audience, often unread but always present, is the Red Hat account team that will eventually see the result of the work indirectly. Both audiences want the same thing from the model: a number that a careful reader cannot dismiss. A plan with a missing line item is dismissible. A plan that prices the move at twenty percent of the actual cost is more dismissible than no plan at all, because it confirms to the account team that the buyer has not yet done the work.1
The destination of choice for most RHEL migrations modeled in the practice is Rocky Linux. The reasons are familiar. Rocky is binary compatible with RHEL at the package level. It is governed by the Rocky Enterprise Software Foundation, which has held the project since the original CentOS announcement in December 2020. It has a paid support option through commercial sponsors and a free path for organisations that can absorb support internally. None of those properties remove the need for an honest cost model. They change which line items are large, not whether they exist.
The four cost categories most plans understate.
Across exit planning engagements observed in the practice, internal cost estimates underprice four categories with such regularity that they can be named in advance. The categories are not exotic. They are the bill the engineering team rarely writes down because it lives in calendar time rather than line items.
The first category is engineering labor per system. The conversion itself, whether run with a community tool such as the migrate2rocky script or scheduled as a fresh build, takes time the plan rarely models honestly. The script run is the small share of the work. The validation, the application team coordination, the change window, the rollback rehearsal, and the post conversion regression each carry hours that the plan tends to absorb into operating capacity that is, in practice, already absorbed.
The second category is the support substitute. Rocky Linux without paid support is a defensible posture in development and test. In production it transfers an obligation. Either the buyer purchases commercial support from a Rocky sponsor such as CIQ or from a third party support firm such as OpenLogic, or the buyer absorbs the obligation internally with named engineers, on call rotations, and an internal advisory cadence on errata. Neither path is free. The model has to price one of them.
The third category is software vendor certification. Many enterprise applications carry certification statements that name RHEL specifically. Oracle Database, SAP NetWeaver, several IBM software lines, and a number of independent software vendor titles support Rocky Linux only by analogy, only on RHEL, or only under a conditional support posture. The migration plan that omits the cost of running the affected workloads on a separately certified base, or the cost of accepting the conditional support posture in writing, omits the most contested line in the bill.2
The fourth category is lifecycle tooling replacement. Buyers who run Satellite, Red Hat Insights, or Smart Management against the existing RHEL estate use functionality that does not move to Rocky Linux for free. The substitute set is real and reachable. It is not zero cost. A plan that prices the destination operating system but assumes the existing management plane continues to function is a plan that does not match the deployment reality after the move.
| Cost category | Frequency understated | Typical understatement |
|---|---|---|
| Engineering labor per system | 12 of 14 | 40% to 70% |
| Support substitute (paid or internal) | 11 of 14 | 25% to 50% |
| Vendor certification on affected apps | 10 of 14 | often not modeled |
| Lifecycle tooling substitute | 9 of 14 | often not modeled |
Where Rocky Linux saves, and where it does not.
The saving is real, and it is bounded. On systems where the RHEL subscription was paying for an entitlement the buyer can absorb internally or substitute commercially at a lower price, the avoided subscription is the net saving minus the substitute cost. On systems where the RHEL subscription was paying for capabilities the substitute does not replace, the saving is smaller than the headline and the residual obligation is larger than the model suggests.
The clearest saving sits in development, test, lab, and disposable workloads. In these tiers the support obligation is light, the certification surface is small, and the management plane is often informal already. Conversion runs cleanly. The saving is close to the full RHEL subscription per system, less the cost of any minimal Rocky support contract or internal absorption. This is also the tier where most observed hybrid postures take their first step. Production stays on RHEL. Development moves to Rocky. The estate shrinks. The remaining renewal carries fewer systems.
The least clear saving sits in production workloads that support a regulated function, run a software vendor application with a named certification on RHEL, or depend on Smart Management or Insights data for an internal audit posture. In these tiers the substitute set is reachable but not free, the certification posture is conditional, and the lifecycle tooling has to be replaced before the move closes. The saving is real but smaller, and the time to realise the saving is longer. A plan that quotes the headline saving across the full estate, without sorting by tier, is not yet a plan.
The same sort applies to OpenShift workloads. The destination set for OpenShift is different from the destination set for RHEL, the cost model is harder to build, and the saving band is narrower. The article on exit planning as a service treats this distinction across product lines. The article on container platform alternatives to OpenShift treats the OpenShift specific case directly.
Migration economics as renewal instrument.
The most consistent observation across the trailing twelve months of exit planning work is that the cost model produces more concessions when it is read than when it is executed. A buyer who arrives at the renewal table with a costed, calendared, sponsor backed Rocky Linux plan modeled across the estate, sorted by tier, and priced including the four categories above, is read by the account team as a buyer who could leave. The renewal proposal that follows is, in observed engagements, not the same proposal that would have followed a renewal cycle opened without the document.
The mechanics are not subtle. The Red Hat account team is incentivised on renewal protection. A buyer with a defensible exit model is a retention case. Retention cases receive concession bands that buyers without an exit model do not. The bands observed in the practice across the eight engagements that ran the documented plan posture and did not execute migration fell between 31% and 58% off the opening renewal number. The bands observed for the five engagements that executed a hybrid migration across part of the estate fell between 22% and 41% on the remaining contract.4
The buyer-side reading of this pattern is that the cost model is not an exit tool. It is a renewal tool that happens to be denominated in migration line items. The plan does not have to be acted on to produce the effect. It has to be readable. A plan that is internally credible and externally legible is the instrument. The discount is the output. The migration is, in most observed cases, optional. The deeper treatment of this dynamic sits on the parent service hub for exit planning, and on the companion note covering the credible migration plan as leverage.
When the model says stay.
An honest cost model produces a recommendation to stay in a non trivial share of engagements. The destination set is real, the saving is bounded, and the saving sometimes does not clear the cost of getting there. The recommendation to stay is itself a buyer-side output. It informs the renewal conversation. A buyer who can demonstrate that the model was run carefully, sorted by tier, priced across all four categories, and still produced a stay recommendation has a different negotiating position than a buyer who never built the model.
The pattern is most visible on estates with heavy regulated workload, deep software vendor certification dependency, or extensive Satellite and Insights use. In these estates the saving net of the support substitute, the certification cost, and the lifecycle tooling substitute is often small. The stay recommendation is the honest output. The renewal conversation that follows is calmer than it would have been without the model, because both sides know the work has been done and the buyer is choosing the relationship. The companion note on when not to migrate off Red Hat sits in this territory directly.
The practice treats Rocky Linux as one of several destinations that may emerge from an honest model, alongside AlmaLinux, Oracle Linux on the equivalent path, and SUSE on the Liberty support path. Each carries its own cost profile and its own concession effect at the renewal table. The companion notes on AlmaLinux migration timeline and risk and Oracle Linux as a RHEL alternative cover the destination level differences. The companion note on how leaving Red Hat changes the current contract covers the renewal effect. The parent practice hub for RHEL across this entire posture sits at the RHEL practice page.
The recurring buyer-side line, across stay and go outcomes alike, is the one worth holding in the room: a credible exit changes the deal even when no one leaves. The cost model is the instrument that makes the exit credible. The decision to act on it is downstream.
Notes & references
- 1. The "twenty percent plan" is a recurring pattern in observed engagements. The internal estimate carries the obvious line items at roughly a fifth of the realistic figure, omits the four categories below, and is read by the account team as evidence that the work has not been done. See internal practice note, "Why an under priced plan is worse than no plan," Issue I, MMXXVI.
- 2. Independent software vendor certification on Rocky Linux is improving in 2026 but remains conditional in several enterprise categories. Oracle Database, SAP, and several IBM software titles support Rocky only by analogy with their RHEL certification or under a posture that conditions support on reproducing the issue on RHEL. The cost model has to price the consequence of that posture for the affected workload, not assume it away.
- 3. Underpricing frequency is reported across the fourteen engagements where the practice rebuilt the internal cost model in detail. The reported understatements are net of the cases where the internal team had already modeled the line item correctly.
- 4. Concession bands reflect the practice's observation across signed renewal contracts in the trailing twelve months. Bands are observations, not promises. Range across the fourteen engagements: 22% to 58% off the opening Red Hat renewal number, with one outlier on the high side after a hybrid migration on more than half of the estate.
- 5. The migrate2rocky utility, maintained by the Rocky Linux project, runs the in place conversion from RHEL or compatible distributions. Equivalent tooling exists in the AlmaLinux ecosystem. The script is a small line in the labor model; the validation, coordination, and rollback rehearsal around it are the larger line.
- 6. The Rocky Enterprise Software Foundation governs the project and the trademark. Commercial support is available through CIQ as the primary sponsor and through third party providers such as OpenLogic. The buyer-side reading is that any production posture on Rocky has to choose one of those paths or absorb the obligation internally with a documented engineering commitment.
Preparing a response? The practice keeps a one-page Red Hat audit response checklist — what to acknowledge, what to preserve, and what not to volunteer in the first fourteen days after the letter arrives.