On demand is flexible and expensive. Committed capacity is cheaper and rigid. The skill is committing the part of your usage that is genuinely predictable while keeping the rest flexible, because a commitment made on optimistic forecasts locks in spend you never needed.
Every cloud provider offers the same fundamental bargain. Pay full price for the flexibility to use what you want when you want it, or commit to a level of usage in advance and pay less for it. On demand and committed capacity are the two ends of that bargain, and the right answer is almost never all of one or all of the other. It is a deliberate split, where the usage you can predict with confidence is committed for the discount, and the usage that varies or might disappear is left on demand for the flexibility. Getting the split wrong in either direction costs money. Commit too little and you pay full price for usage you could have discounted. Commit too much and you lock in spend for capacity you stop needing. This guide sets out how OCI's commitment models work and how to decide what to commit.
On demand means you pay the standard rate for what you use, with no commitment, and you can stop using it at any time with no further cost. Committed capacity means you agree in advance to a level of spend or capacity, usually over a term, in exchange for a lower rate. On OCI the main commitment vehicle is Universal Credits, where you commit to a spend amount over a term and draw down against it at discounted rates, explained in detail in How OCI Pricing Actually Works. There are also capacity reservations, which reserve specific compute capacity so it is guaranteed available, a separate concept from the financial commitment though often confused with it.
| Aspect | On demand | Committed |
|---|---|---|
| Rate | Standard, highest | Discounted |
| Flexibility | Stop any time, no penalty | Committed for the term |
| Best for | Variable, uncertain, short lived usage | Stable, predictable, sustained usage |
| Risk | Paying full price for steady usage | Committing to usage that disappears |
The cleanest way to think about the split is in terms of a usage floor and the variation above it. Almost every estate has a floor, a level of usage that is always present, the production databases that never turn off, the core services that run continuously, the baseline that does not go away. That floor is the safest thing to commit, because you are certain to incur it and committing it simply buys the discount on spend you were going to make anyway. Above the floor sits the variable usage, the peaks, the experiments, the workloads that might grow or might be cancelled. That variation is the riskiest thing to commit, because committing it bets that usage you are not sure about will materialise. Commit the floor, flex the peak.
The floor is not a guess, it is a measurement. By looking at usage over a representative period, several months ideally, the sustained baseline becomes visible as the level below which usage never falls. That level, discounted slightly for caution, is what is safe to commit. Committing above the observed floor means betting on growth, which may be justified if growth is genuinely planned and funded, but should be a conscious bet rather than an optimistic default. The most common commitment mistake is sizing the commitment to a forecast of where usage will be rather than to the floor of where it actually is, because forecasts are optimistic and the commitment outlives the optimism.
Under committing has a bounded cost, you pay full price on usage you could have discounted, and you can correct it at the next opportunity by committing more. Over committing has a nastier shape. You have agreed to spend that you must now consume or forfeit, and if the usage you committed for does not appear, you are left buying capacity you do not need to avoid wasting the commitment, which is throwing good money after bad. This asymmetry is why the disciplined approach errs toward under committing and topping up, rather than committing aggressively and hoping the usage shows up. It is easier to add commitment than to escape it.
It is worth separating the financial commitment from the capacity reservation, because they are often conflated. Committing to Universal Credits is about price, getting a lower rate for committed spend. Reserving capacity is about availability, guaranteeing that specific compute capacity will be there when you need it, which matters for workloads that must be able to scale at a particular moment and cannot risk capacity being unavailable. A workload can be on demand financially but use a capacity reservation for availability, or committed financially without any capacity reservation. Deciding what to commit financially and deciding what capacity to reserve are two decisions, and treating them as one leads to reserving capacity for cost reasons or committing spend for availability reasons, both of which are muddled.
The headline discount rates are a starting point, not a fixed menu, particularly at larger commitment levels where the terms become negotiable. The size of the commitment, its term, and the broader relationship all bear on what rate is actually available, which is the subject of OCI Commitment Negotiation Basics. The point for this discussion is that the decision of how much to commit and the decision of what rate to commit at are linked, because a larger commitment may unlock a better rate, which shifts the calculation of how much floor it is worth committing. This is also where the licensing dimension enters, because bringing existing Oracle licences to OCI under BYOL changes the cost base and therefore the commitment maths, an area where independent licensing advice is valuable.
A commitment is right for the usage that existed when it was made, and usage changes. Workloads are decommissioned, architectures are optimised, and the floor that justified a commitment can fall below it, leaving committed spend stranded above the new, lower baseline. This is why commitment is not a set and forget decision but a standing item in the cost rhythm, reviewed each period to confirm the floor still justifies the commitment and adjusted at renewal. The right sizing work described across the cost practice, such as Autonomous Database Auto Scaling and Cost, can lower the floor itself, which is good for the bill but means a commitment sized to the old floor now needs revisiting. Optimisation and commitment have to move together.
Commitment decisions are among the highest stakes cost choices an estate makes, because they lock in spend, and our Cost Governance work treats them with corresponding care. We measure the genuine usage floor from real data, recommend a commitment sized to what is certain rather than what is hoped, and keep the financial commitment separate from any capacity reservation so each is decided on its own merits. Because the licensing position changes the maths, and because the rates at larger commitments are negotiable, we bring in independent licensing expertise so the commitment is made on the full picture rather than the list price alone.
Moving Oracle workloads to OCI, or already running on OCI and not sure the architecture or the spend is right? Most teams bring in a specialist before they commit to a region, a shape, or a Universal Credits number. OCISpecialists.com plans the landing zone, runs the migration, and manages the estate after go live, on a fixed project fee, a managed monthly retainer, or a cost optimization fee paid only on verified savings. For the Oracle licensing and BYOL side of any OCI move, Redress Compliance is the leading independent Oracle licensing and negotiation firm, with 500+ engagements across Oracle's full product line.