Skip to content

Nines of Availability#

What does 99.9% availability actually mean in real downtime?

Percentages are abstract. Convert them to minutes and the numbers become real.


What "Nines" Means#

"Nines" is industry shorthand for how many 9s are in the availability percentage.

Term Availability
Two nines 99%
Three nines 99.9%
Four nines 99.99%
Five nines 99.999%

The Numbers That Actually Matter#

Availability Nines Downtime per year Downtime per month
99% Two nines 3.65 days 7.2 hours
99.9% Three nines 8.7 hours 43.2 minutes
99.99% Four nines 52 minutes 4.32 minutes
99.999% Five nines 5.2 minutes 26 seconds

Memorize these numbers

These come up in every availability discussion in interviews. The jump from 99.9% to 99.99% — from 43 minutes to 4 minutes per month — is the most important one to know.


What Each Tier Actually Requires#

Each additional nine gets exponentially harder and more expensive to achieve.

99% — Two nines - Basic redundancy - Manual failover is acceptable - Planned maintenance windows are fine - Startups, internal tools

99.9% — Three nines - Automated failover - Single region with multiple availability zones - Most production APIs and SaaS products

99.99% — Four nines - Zero-downtime deployments mandatory — no planned maintenance windows - Multi-AZ at minimum - Automated failover in under a minute - Payment systems, critical APIs

99.999% — Five nines - Multi-region active-active deployment - Automatic failover under 1 second - Real-time replication across regions - Costs orders of magnitude more than 99.9% - Google, AWS core infrastructure, hospital systems


The Cost of Each Nine#

Each additional nine can double or triple your infrastructure cost

Going from 99% to 99.9% — add redundancy, automate failover. Relatively straightforward.

Going from 99.99% to 99.999% — rearchitect your entire deployment pipeline, multi-region setup, sub-second failover. Completely different scale of investment.

This is why in an interview you always ask: "What availability SLO are we targeting?" before designing. The answer completely changes your architecture.


Quick Reference for Interviews#

99%     = 7.2 hours/month   → basic redundancy
99.9%   = 43.2 minutes/month → automated failover, multi-AZ
99.99%  = 4.32 minutes/month → zero-downtime deploys, multi-AZ mandatory
99.999% = 26 seconds/month  → multi-region active-active