Survive an AZ outage

advanced

Scenario & brief

This service runs two instances across two AZs with a Multi-AZ database — but if an entire AZ goes dark, the survivors can't absorb full peak, and the database has no read replica to fall back on.

Targets: 99.99% availability, critical durability, p99 ≤ 160 ms, under $1,300/mo.

Make it genuinely AZ-outage-ready: spread across three AZs with enough instances that any two can still serve peak, add a read replica alongside Multi-AZ for the data tier, and use a cache to keep the database comfortable.

5,000 rps peakp99 ≤ 160ms99.99% availdurability: criticalbudget $1,300/mo

App Servers

Compute

Instance class

Scaling

Instances (static)

2instances

Availability Zones

Purchasing

System health

Erupting · SLA breach

/ 100

Score

SLA not met yet

Monthly cost

$653

Budget $1,300/mo · within budget

Metrics

Capacity34

Availability90

Durability25

Cost efficiency80

Requirements

Peak capacity — 3600 rps compute · 1700 rps db (need ≥ 5000 rps)
p99 latency — ~194 ms (need ≤ 160 ms)
Availability — 99.95% (need ≥ 99.99%)
Durability — at risk (need redundancy + backups)
Budget — $653/mo (need ≤ $1300/mo)

Advisor

Compute tops out at ~3600 rps but peak demand is 5000 rps — requests will queue.
The database is saturated at peak — add a cache to shed read load, or scale it up.
Enable automated backups to satisfy the durability requirement.

Discussion

No comments yet. Be the first to start the discussion.

For learning purposes only. Costs and capacities are illustrative, not live AWS prices. Not affiliated with or endorsed by Amazon Web Services.