Composite Availability Calculator
Inputs
| Arrangement | Series |
|---|---|
| Components | Two |
| Component 1 | 99.9 % |
| Component 2 | 99.9 % |
| Component 3 | 99.9 % |
Composite Availability Calculator
Combine the availability of two or three components in series or in parallel (redundant) using reliability block diagrams. Get the composite system availability and the resulting downtime per year.
Inputs
Component Availability
Results
Enter a value to see results.
System Availability
Expected Downtime
Composite Availability
Availability is the long-run fraction of time a component is operational. A single server, link, or service has its own availability — often quoted in "nines," such as 99.9 % (three nines). Real systems, though, are built from many such parts, and the availability that users actually experience depends on how those parts are wired together. Composite availability is the availability of the whole system, computed from the availability of each component and the topology of the reliability block diagram that connects them.
Two arrangements cover most cases. In series, every component must work for the system to work. In parallel, the components are redundant: any one of them keeps the system running. Series arrangements erode availability; parallel arrangements build it up. This calculator combines two or three components in either arrangement and reports the composite availability together with the downtime it implies over a year.
Formula
Let each component have availability , a number between 0 and 1.
For components in series, the system is available only when all of them are available. Assuming independent failures, the availabilities multiply:
A=i∏Ai=A1×A2×⋯The result is always lower than the least-available component — every extra part in series is another thing that can fail.
For components in parallel (redundant), the system is unavailable only when every path is unavailable at the same time. It is easier to reason about the unavailability of each part, since those are what must all coincide:
A=1−i∏(1−Ai)=1−(1−A1)(1−A2)⋯The result is always higher than the best component, because a redundant copy can cover for a failed one.
Once the composite availability is known, the expected downtime over a year is the unavailable fraction multiplied by the length of a year:
Dyear=(1−A)×525,600 min=(1−A)×8,760 hWorked Example
Take two components, each with 99.9 % availability (), and combine them both ways.
In series:
A=A1×A2=0.999×0.999=0.998001=99.8001%Placing the two components in series lowers the availability from 99.9 % to 99.8 %, roughly doubling the expected downtime. The system now spends about 17.5 hours a year unavailable.
In parallel:
A=1−(1−A1)(1−A2)=1−(0.001)(0.001)=1−0.000001=99.9999%Making the two components redundant raises the availability from three nines to six, cutting the expected downtime from hours to about half a minute a year. The same two parts, arranged differently, differ in downtime by a factor of roughly 2,000.
Series Removes Nines, Parallel Adds Them
Because series availabilities multiply, stringing components together steadily removes nines. Ten identical services at 99.9 % in a single series chain give — the chain has fallen to about two nines even though every link is at three. This is why a request path that touches many services can be far less reliable than any one service on it.
Redundancy works in the opposite direction. Each parallel copy multiplies the already-small unavailability by another small number, so nines accumulate quickly:
| Redundant copies (each 99.9 %) | System availability | Downtime per year |
|---|---|---|
| 1 | 99.9 % | ~8.76 hours |
| 2 | 99.9999 % | ~31.5 seconds |
| 3 | 99.9999999 % | ~0.03 seconds |
The steep climb is the central appeal of redundant design: doubling the hardware turns three nines into six.
Nines and Downtime
The downtime-per-year figure is usually the most tangible way to read an availability number. Each additional nine cuts the allowed downtime by about a factor of ten:
| Availability | Downtime per year |
|---|---|
| 99 % (two nines) | ~3.65 days |
| 99.9 % (three nines) | ~8.76 hours |
| 99.99 % (four nines) | ~52.6 minutes |
| 99.999 % (five nines) | ~5.26 minutes |
| 99.9999 % (six nines) | ~31.5 seconds |
Composing components in series moves down this table; adding redundancy in parallel moves up it.
Assumptions and Limits
The multiplication rules assume that component failures are independent — that one part going down does not make another more likely to go down. Redundancy only delivers its dramatic gains when the copies truly fail independently. Shared power, a common network, a single load balancer in front of a redundant pair, or a software bug present in every copy all introduce correlated failures that the simple formula does not capture; the real availability of such a system is lower than the parallel result suggests.
The calculation also treats availability as a steady long-run average and ignores repair time, failover delay, and planned maintenance windows unless those are already folded into each component's availability figure. For a first-order estimate of how an arrangement of components will behave, though, the series and parallel rules are the standard tools, and they make the trade-off between adding parts and adding redundancy easy to see.
For the downtime figures behind a single availability target, see Uptime SLA Calculator.
Frequently Asked Questions (FAQ)
What is composite availability?
Composite availability is the availability of a system built from several components, worked out from the availability of each part and the way the parts are arranged. Availability is the long-run fraction of time a component is operational — 99.9 % availability means the component is up 99.9 % of the time. A reliability block diagram represents the system as blocks connected in series, in parallel, or in a mix of the two, and the composite availability follows from how those blocks combine.
How do series and parallel differ?
In a series arrangement every component must be working for the system to work, so a failure anywhere brings the whole system down. The availabilities multiply, and the result is always lower than the weakest component. In a parallel (redundant) arrangement the components back each other up: the system stays up as long as at least one path is working, and it is down only when all of them fail at once. There the unavailabilities — the 1 − A terms — multiply, so the result is higher than the best component.
Why does redundancy improve availability so much?
Because independent failures rarely coincide. If one component is down 0.1 % of the time, two identical components in parallel are both down only 0.1 % × 0.1 % = 0.0001 % of the time, turning three nines into six. Each redundant copy multiplies the already-small unavailability by another small number, so availability climbs steeply. The gain assumes the components fail independently; shared power, a common dependency, or a single switch in front of the pair introduces correlated failures that redundancy cannot remove.
How do the "nines" relate to downtime?
Each additional nine cuts the allowed downtime by roughly a factor of ten. Two nines (99 %) permits about 3.65 days of downtime a year; three nines (99.9 %) about 8.76 hours; four nines (99.99 %) about 52.6 minutes; five nines (99.999 %) about 5.26 minutes. Combining components in series removes nines, while adding redundancy in parallel earns them back, which is why the downtime-per-year figure is often the most tangible way to read a composite availability result.
Recommended Next
Uptime SLA Calculator
Convert a service-level agreement availability percentage into allowable downtime per year, month, week, and day.