Availability is a game of nines. As you add a nine to the end of your percentage, you are reducing your allowable downtime by a factor of 10. This requires exponentially more complex infrastructure, redundancy, and automated recovery systems.
The Downtime Math
Let's look at what these numbers actually mean in terms of time your system is allowed to be dead to the world.
| Availability Tier | Downtime / Year | Downtime / Month | Downtime / Week |
|---|---|---|---|
| 99.9% ("Three Nines") | 8.76 hours | 43.8 minutes | 10.1 minutes |
| 99.99% ("Four Nines") | 52.6 minutes | 4.38 minutes | 1.01 minutes |
| 99.999% ("Five Nines") | 5.26 minutes | 26 seconds | 6 seconds |
The Operational Gap
Three Nines (99.9%) is the industry standard for most commercial SaaS applications. It allows for nearly 45 minutes of downtime a month. This means you can survive a bad deployment, roll back, and maybe even have a small database outage, all while staying within your SLA.
Four Nines (99.99%) is a different beast. You have 4 minutes a month. That is not enough time for a human to wake up, check PagerDuty, open a laptop, and debug. Four nines requires automated failover, active-active redundancy, and zero-downtime deployments. It is the realm of core infrastructure providers like AWS S3 or Google Spanner.
Recover your SLA credits
Stop leaving money on the table. Breachr automates the evidence gathering and negotiation process for you.