Coverage
Reserved Instances only save money on the usage they cover.
Coverage is the one number that tells you if a fleet is overpaying — in either direction. Too little, and steady usage keeps paying the on-demand premium. Too much, and reservations sit on capacity that isn't there anymore. Here's what coverage measures, why both sides cost real money, and how to set a target that scales with you.
First, the metric
Coverage is % of reservable running capacity matched to a reservation.
Take everything currently running — RDS, ElastiCache, and EC2 usage that AWS lets you reserve — and ask how much of it is actually matched to a Reserved Instance you own, right now. That share is coverage. Usage above 100% coverage doesn't exist; usage below it is paying the standard on-demand rate for something that could be discounted.
Discount ceilings per AWS published pricing; your actual rate depends on instance class, term, and payment option.
Not every running unit is reservable — short-lived or serverless usage doesn't qualify. Coverage is measured only against the reservable slice of your fleet, so it stays an honest read on how well your reservations are doing their job, not diluted by usage that was never a candidate.
What it looks like
One fleet, three states: reserved, on-demand, expiring.
This fleet sits at 60% — comfortable, but not the goal by itself. The outlined units are the real question: are they variable usage that should stay on-demand, or steady usage that's simply never been reserved? The amber units are the renewal cliff from the renewals guide — still covered today, on-demand rate starting the moment the term ends.
The money is on both sides
Under-coverage and over-coverage are both a leak — just in opposite directions.
Steady usage, paying the on-demand rate
Any running instance that's on-demand every hour, month after month, is leaving the Reserved Instance discount on the table — up to 72% on EC2, 69% on RDS, 55% on ElastiCache. It doesn't show up as an incident; it's just the standard rate, quietly, for usage that was always going to be there.
Reservations covering nothing real
A reservation whose instance was resized, moved, or terminated keeps billing at the reserved rate for the rest of the term — it's dead weight, not savings. And reserving 100% of usage that is itself variable means paying for idle capacity on every day you don't hit the peak.
Setting the target
A target is a floor plus headroom — not a fixed percentage.
Base coverage to the steady floor
Find the minimum amount of an instance class that's running concurrently, day-round — that floor is what should be reserved first. It's the usage least likely to ever go away.
Leave headroom for what's variable
Bursty or seasonal capacity above the floor stays on-demand on purpose. Reserving it commits you to paying for a peak on the days you don't hit it.
Pick the term by how stable the family is
An instance family with years on your bill is a safer 3-year bet; something new, or due to be resized, costs less certainty on a 1-year term.
Recompute every term, from current usage
Set the target from what's running today, not from what the last purchase covered. Fleets change shape between terms — the target should too.
How this scales with you
The target doesn't change with fleet size. Keeping it does.
A healthy target is easy to reach by eye — most of a small, steady fleet can be reserved outright. The risk isn't complexity, it's forgetting to revisit the target as the fleet grows past the point you can eyeball it.
The floor moves every month as instances launch, resize, and get replaced. Without a live view, coverage quietly drifts away from the target in whichever direction nobody's watching — under, or over.
Coverage has to be one number across every account and cloud, not a report stitched together once a quarter. At this scale, a few points of drift either way is real money, every month it goes unnoticed.
Ten reservations or ten thousand, the math behind the target is the same — only the effort to keep tracking it by hand changes. See pricing: one flat price at every size, never a cut of what you save.
Where Reserver fits
One coverage view, kept current automatically.
Reserver connects to your AWS accounts read-only, scans every reservation you own alongside what's actually running, and computes coverage against your fleet's own steady floor — not a guess. The matcher flags reservations with nothing left to cover; recommendations size the next purchase to close the real gap, not round up to 100%; expiry radar keeps the amber units in view before they turn red. See features, how it works, the security model behind the read-only connection, and the savings calculator to estimate the gap against your own spend — or, if your fleet also carries Savings Plans, the RI vs Savings Plans guide for how the two compare.
What's a good coverage target?
There's no universal number — it's the size of your steady, always-on floor as a share of total running capacity, plus a little headroom for anything variable. A fleet that's mostly steady state might target well above 80%; a bursty one might target closer to 50%. The right target comes from your own usage pattern, not a rule of thumb.
Can I have too much coverage?
Yes. A reservation only saves money while it's covering real, matching usage. Reserve past your steady floor — to cover a peak that doesn't repeat, or an instance that gets resized or terminated — and you're paying for a discount on capacity that no longer exists, for the rest of the term.
Does coverage include Savings Plans?
Not today. Reserver tracks Reserved Instance coverage across RDS, ElastiCache, and EC2. Savings Plans are on the roadmap, not in v1 — if you're mixing both, RI coverage is only part of your picture for now. See the Reserved Instances vs Savings Plans guide for how the two compare.
What about instances that come and go?
They're exactly why coverage needs recomputing regularly rather than set once. An instance that's resized or replaced can leave a reservation with nothing left to match — the fix isn't to avoid reserving it, it's to see the mismatch quickly and act on current usage instead of last term's.
Stop overpaying for a steady fleet.
Reserver is in early access. Join the list and we'll onboard you as capacity opens up — first recommendation within minutes of connecting.