AWS Reserved Instances cut EC2 cost by up to 72 percent versus on-demand for a 1 or 3 year commitment, with Standard RIs reaching the deepest discount and Convertible RIs trading about 10 percentage points of savings for the right to change instance family; for most workloads Savings Plans now match RI economics with more flexibility. Reserved Instances were the original AWS commitment discount, and they still deliver the largest per-instance savings AWS offers, but the introduction of Savings Plans changed when an RI is the right tool. This page covers the mechanics, the trade-offs, and the decision.
How Reserved Instances work
A Reserved Instance is not a reserved server; it is a billing discount applied to matching on-demand usage in exchange for a 1 or 3 year commitment to a specific instance configuration. You commit to an instance family, size, region, and tenancy, and AWS applies the RI's lower rate to any running instance that matches. The discount scales with term length and payment option: a 3-year all-upfront Standard RI reaches the maximum 72 percent off on-demand, while a 1-year no-upfront Convertible RI sits closer to 30 percent. The commitment is financial, not capacity, so an unused RI still bills.
Standard versus Convertible
| Attribute | Standard RI | Convertible RI |
|---|---|---|
| Max discount (3yr, all-upfront) | Up to 72% | Up to 62% |
| Change instance family | No | Yes, via exchange |
| Change size within family | Yes (Linux, regional) | Yes |
| Sell on Marketplace | Yes | No |
| Best for | Stable, known workloads | Evolving workloads |
Standard RIs give the deepest discount but lock you to an instance family for the term, with the escape hatch of selling unwanted Standard RIs on the Reserved Instance Marketplace. Convertible RIs sacrifice roughly 10 points of discount for the right to exchange into a different family as your architecture changes, but they cannot be resold. The choice is a bet on how stable your instance choices are over the commitment term.
Payment options
| Payment option | Upfront | Discount depth | Cash-flow profile |
|---|---|---|---|
| All upfront | Full term paid day one | Deepest | Largest upfront outlay |
| Partial upfront | Part now, rest monthly | Near-deepest | Balanced |
| No upfront | Nothing now, monthly | Shallowest | Smoothest |
The discount difference between all-upfront and no-upfront is usually only 2 to 4 percentage points, so the payment choice is more a cash-flow decision than a savings decision. For most enterprises the small extra discount of all-upfront does not justify tying up the capital, and partial or no-upfront preserves flexibility at almost the same rate.
Reserved Instances versus Savings Plans
Savings Plans deliver the same compute discount depth as RIs but commit to a dollar-per-hour spend rather than a specific instance configuration. Compute Savings Plans apply automatically across instance families, sizes, regions, and even Fargate and Lambda, which removes the management burden of matching RIs to usage. For most workloads this flexibility makes Savings Plans the better default, and AWS has steered new commitments toward them. Reserved Instances retain two real advantages: zonal RIs can reserve capacity in a specific Availability Zone, which Savings Plans cannot, and Standard RIs can be resold on the Marketplace to recover value from an over-commitment. The full comparison is in our Savings Plans versus Reserved Instances analysis and the Savings Plans guide.
The coverage-efficiency lever: The expensive mistake is buying one commitment type to cover an entire fleet. Steady, capacity-sensitive workloads in a fixed AZ are best on zonal Standard RIs; diverse, evolving compute is best on Compute Savings Plans; and a baseline of predictable spend can take the deepest 3-year commitment while a variable layer stays on-demand. Blending the instruments to match each workload's stability, rather than standardizing on one, typically improves coverage efficiency by 15 to 20 percent. Model the blend, do not default to it.
The Reserved Instance Marketplace
Standard RIs that no longer match your usage can be sold on the AWS Reserved Instance Marketplace, recovering a portion of an over-commitment that would otherwise be sunk. The resale market is a meaningful risk reducer for Standard RIs and a reason some buyers prefer them over Convertibles despite the lost flexibility. Convertible RIs cannot be sold, only exchanged, so an over-bought Convertible position can only be reshaped, not unwound for cash.
Capacity reservation versus the discount
A point of frequent confusion is that a Reserved Instance does not, by default, guarantee capacity. A regional RI is purely a billing discount and reserves nothing; if a region is constrained, a regional RI does not promise you an instance. A zonal RI, scoped to a specific Availability Zone, does reserve capacity in that zone. For workloads that must be able to launch in a specific zone during a regional event, the zonal RI is the only commitment instrument that delivers both the discount and a capacity guarantee, which Savings Plans cannot match. Where capacity assurance is not required, the regional RI's flexibility across zones is the better default.
Utilization and waste
The discount only materializes if the RI is matched by running usage, so utilization is where RI value is won or lost. An RI that sits unmatched because the workload it covered was retired or resized is pure waste, billing at the committed rate for nothing. Mature RI programs monitor utilization continuously, exchange or resell positions that drift out of use, and size commitments to the stable baseline of demand rather than the peak. A common discipline is to cover only the bottom 60 to 80 percent of predictable demand with commitments and leave the variable top layer on-demand, which keeps utilization high and avoids paying for reservations that idle.
| Coverage layer | Demand profile | Best instrument |
|---|---|---|
| Baseline (always on) | Stable 24/7 | 3-year Standard RI or Savings Plan |
| Predictable variable | Daily or weekly pattern | 1-year commitment or Compute SP |
| Spiky or experimental | Unpredictable | On-demand or Spot |
| Zone-critical | Must launch in set AZ | Zonal Standard RI |
Instance size flexibility
Regional Linux RIs carry instance size flexibility, meaning the discount applies across sizes within the same family and generation in proportion to a normalization factor. A reservation for one large instance can cover two mediums or four smalls of the same family, which materially reduces the risk of buying the wrong size. This flexibility narrows the practical gap between Standard and Convertible RIs for many workloads, because it lets a Standard RI flex across sizes even though it cannot change family. Understanding the normalization factor is what lets a buyer commit confidently to Standard RIs without fearing a resize will strand the discount.
A worked blended strategy
The strongest commitment programs do not pick one instrument; they layer several to match the shape of demand. Consider a fleet with a stable 24/7 baseline of $400,000 a month in compute, a predictable daytime variable layer of $150,000, and a spiky experimental layer of $80,000. The efficient structure covers the baseline with a 3-year Compute Savings Plan or Standard RIs at the deepest discount, covers the predictable variable layer with a 1-year commitment that can be renewed as the pattern confirms, and leaves the spiky layer on-demand or on Spot where interruption is acceptable. Zone-critical components within the baseline take zonal Standard RIs for the capacity guarantee. The result captures deep discounts on the spend that justifies a long commitment while keeping flexibility on the spend that does not.
| Layer | Monthly spend | Instrument | Approx. discount |
|---|---|---|---|
| Stable baseline | $400,000 | 3-year Savings Plan / Standard RI | Up to 72% |
| Predictable variable | $150,000 | 1-year Compute Savings Plan | 40 to 54% |
| Spiky experimental | $80,000 | On-demand / Spot | 0 to 90% (Spot) |
Run against a self-managed single-instrument approach, this layered structure typically improves coverage efficiency by 15 to 20 percent, because it stops paying for reservations that idle and stops leaving stable spend uncommitted. Modeling the layers against real utilization data is the work; defaulting to one instrument across the fleet is the mistake.
Frequently asked questions
Are Reserved Instances still worth buying over Savings Plans?
For most workloads Savings Plans are the better default because of their flexibility. RIs remain worth buying for two specific needs: zonal capacity reservation, which Savings Plans cannot provide, and the ability to resell a Standard RI on the Marketplace to recover an over-commitment.
Does a Reserved Instance guarantee I can launch an instance?
Only a zonal RI does. A regional RI is purely a billing discount and reserves no capacity; if you need a capacity guarantee in a specific Availability Zone, you must buy a zonal RI.
What payment option should I choose?
The discount gap between all-upfront and no-upfront is usually only 2 to 4 points, so for most enterprises partial or no-upfront is the right choice, preserving capital at almost the same rate.
Monitoring, automation, and accounting
A commitment program is not a one-time purchase but an ongoing position that needs management. Utilization and coverage should be monitored continuously, because workloads change and a commitment that fit last quarter can drift out of alignment this quarter, leaving discount unused or spend uncovered. Mature programs automate this monitoring, flagging under-utilized reservations for exchange or resale and identifying uncovered baseline spend that should be committed. The accounting treatment also matters for larger buyers: all-upfront Reserved Instances are a prepaid asset amortized across the term, while no-upfront commitments are recognized as incurred, and the choice interacts with how finance prefers to treat the spend. Aligning the payment option to both the cash-flow preference and the accounting treatment, rather than defaulting to whichever AWS proposes, is a small decision that compounds across a large fleet. The combination of continuous monitoring, automated rebalancing, and a deliberate accounting choice is what turns a static commitment into a managed position that keeps delivering its modeled savings over the full term.
Recommendations
Default new compute commitments to Compute Savings Plans for the flexibility, reserve Standard RIs for stable workloads that need zonal capacity reservation or that you may want to resell, and use Convertible RIs only where you need instance-family flexibility that Savings Plans cannot provide. Layer the committed-spend EDP discount on top, and remember that every committed dollar also reduces the base your enterprise support cost is calculated on. The committed-spend interaction is covered in our committed spend discount guide and the EDP pillar; for engagement, see our software licensing advisory service and the AWS vendor hub.