Composite SLA Calculator
Staff email does not depend on one service. It needs sign-in, DNS, the mail service and your own network, all at once, and the availability of the whole chain is lower than any one of them. Build your chain, mark which parts are redundant, and see the composite availability, the expected downtime and the weakest link. When the Microsoft 365 outage that began on 12 November 2026 hit several services together, it also showed the limit of this maths: it assumes failures are independent. Read what we know in our Microsoft 365 outage briefing.
1. Build the chain
Items run in series: the service works only if every item works. Make an item a redundant group when any one of its members is enough (two internet links, two DNS providers). Enter each availability as a percentage: your provider's SLA figure or your own measured uptime.
Item availability 99.95% · about 21.6 min down per month
Item availability 99.99% · about 4.3 min down per month
Item availability 99.9% · about 43.2 min down per month
Item availability 99.5% · about 3 h 36 min down per month· weakest link
28 to 31. Used 30 days = 43,200 minutes. A year is counted as 365 days.
Example figure you can use: Microsoft's SLA sets a 99.9% monthly uptime commitment for Exchange Online (also Teams, SharePoint Online and OneDrive for work or school). It is the level below which a service credit applies, not a promise of uptime.
Formula: a1 × a2 × a3 × a4 = 99.3409%. Downtime = (1 − A) × minutes in the period. Expected values assume independent failures; a real month can be better or much worse.
What if you add redundancy?
The copy is assumed to have the same availability and to fail independently (a different provider, path or site). A second copy from the same provider, on the same path, gains little. The weakest link then becomes Exchange Online.
If failures line up
| Assumption | Availability | Per month |
|---|---|---|
| Independent failures (headline) | 99.3409% | 4 h 45 min |
| Worst case: outages never overlap | 99.34% | 4 h 45 min |
| Best case: outages always overlap | 99.5% | 3 h 36 min |
The worst and best cases are the limits for any pattern of overlap, given your inputs. When a redundant group's members share a provider, a region or a sign-in system, the “fails together” line is the honest one: the redundancy adds nothing.
The independence assumption
Multiplying availabilities is only right when components fail independently. A shared-provider outage breaks that. The Microsoft 365 outage that began on 12 November 2026 affected Exchange Online, Outlook, Teams, SharePoint, Copilot and Defender XDR together, per Microsoft and press reports (our briefing). Treating those services as independent items, or as backups for each other, would have overstated availability. Score how much of your stack depends on one provider with the SaaS Concentration Risk Scorer.
An SLA percentage is also not a forecast. Microsoft's 99.9% for Exchange Online is the monthly uptime below which a service credit applies under the Service Level Agreement for Microsoft Online Services (WW), November 1, 2026 (read 2026-11-13). Use measured uptime where you have it. To turn one SLA figure into allowed downtime, use the SLA Uptime Calculator.
The chain, the composite, the what-if and the overlap limits, ready for a resilience review or a design note.
TL;DRWhat this calculator does
4 points
- Build a chain of the components a service needs (identity, DNS, a SaaS app, your network) and give each one an availability percentage. Any item can be a redundant group where one member is enough.
- You get the composite availability, expected downtime per month (you set the month length) and per year, the weakest link, and what happens if you add an independent copy of any item.
- Series: A = a1 × a2 × … × an. Redundant group: A = 1 − (1 − a1) × (1 − a2) × …. Both assume failures are independent, so the tool also shows the result when redundant members fail together and the worst and best cases for any overlap.
- Free to use, signed out too. Everything runs in your browser and nothing is sent to us. Markdown and CSV exports are available with a paid plan.
Worked example: email for staff
Four items in series: identity (sign-in) 99.95%, DNS 99.99%, Exchange Online 99.9% and your network on one internet link 99.5%. Only the Exchange Online figure comes from a provider (Microsoft's SLA commitment); the other three are example inputs we chose to show the maths, not any provider's figure.
- One internet link: 0.9995 × 0.9999 × 0.999 × 0.995 = 99.3409%, about 4 h 45 min of downtime in a 30-day month and 2 d 10 h a year. The network is the weakest link.
- Add a second, independent 99.5% link: the network becomes 1 − 0.005 × 0.005 = 99.9975%, and the chain becomes 99.8376%, about 1 h 10 min a month and 14 h 14 min a year. Exchange Online is now the weakest link.
- If both links fail together (same carrier, same street cable), the pair is only as good as one link and the chain is back to 99.3409%.
Load both versions with the example buttons in the calculator. All results are estimates, not advice.
Frequently asked questions
How do you calculate a composite SLA?
For components that are all needed (in series), multiply their availabilities: 99.9% × 99.5% = 0.999 × 0.995 = 99.4005%. For a redundant group where any one member is enough (in parallel), multiply the chances that each member is down and subtract from 1: two independent 99.5% links give 1 − 0.005 × 0.005 = 99.9975%. Combine the two rules for a chain that contains redundant groups.
Why is the composite availability lower than every component?
In a series chain every component has to be up at the same time, so each one adds its own downtime. Four components that each look good can together fall below the weakest of them. That is why the weakest link matters most, and why adding a dependency always costs some availability.
How much downtime is 99.9% availability?
In a 30-day month (43,200 minutes) 99.9% allows 43.2 minutes, 99% allows 432 minutes (7.2 hours) and 95% allows 2,160 minutes (36 hours). In a 28-day February 99.9% allows 40.32 minutes. Over a 365-day year 99.9% is about 8.8 hours.
Does adding a second provider always improve availability?
Only if the two fail independently. Two links from the same carrier, two services on the same cloud region, or two apps that share one sign-in system tend to fail together, and then the group is only as available as its best member. The calculator shows that "fails together" figure next to the independent one.
Is Microsoft 365 99.9% uptime a guarantee?
No. Microsoft's Service Level Agreement for Microsoft Online Services sets a 99.9% monthly uptime commitment for Exchange Online, Microsoft Teams, SharePoint Online and OneDrive for work or school. Below it, a customer can claim a service credit (25% of that service's monthly fees below 99.9%, 50% below 99%, 100% below 95%). It is a financial remedy, not a promise that the service will be up.
Is my data sent anywhere?
No. The calculation runs in your browser. Your chain is saved in this browser's local storage so you can come back to it, and clearing site data removes it.
More security tools
The tools people use most on PlayCISO.
Catch injected instructions hiding in content your AI agent should only read.
Score your readiness against the first international standard for AI security threats.
Critical CVEs across 21 major vendors, month by month — and what the 2026 surge means.
Build an ordered ruleset, replay labelled packets and catch shadowed and any-to-any rules.
Self-assess all six Functions, from Govern to Recover, and see your weakest one.
Search 172 threat actors across Microsoft, CrowdStrike, Mandiant and community names.