N+1 Redundancy

Overview

N+1 redundancy means having N units that carry the load plus one spare. It is the baseline availability design for fleets, power supplies, and storage arrays.

How it works

Work out how many units peak load needs when each runs at a safe utilization, then add one.

peak load            9000 rps
safe per node        1500 rps (70 percent of max)
N = 9000 / 1500      = 6
deploy N+1           = 7 nodes
after one failure    6 nodes at 1500 rps each

For zones the math changes. With three zones and one allowed to fail, each zone must be able to take half of the load, so total capacity is 150 percent of peak. With more zones the overhead drops, for example four zones need about 133 percent.

Pitfalls

• Teams measure utilization with all units healthy, so alerts miss the fact that one failure would breach limits. Alert on N-1 utilization.
• Placing the spare in the same rack or zone defeats the purpose.
• During a deployment one unit is already out of rotation, so a rolling update plus one failure needs N+2.
• Load balancers may not redistribute evenly and overload a few survivors.
• Growth silently consumes the spare, so review capacity regularly.

Key takeaways

• Size N from peak load at a safe utilization.
• Spread units across independent failure domains.
• Monitor headroom as if one unit were already lost.
• Use N+2 when deployments or correlated failures overlap.