Global-scale datastore

expert

Scenario & brief

A key-value platform serves 20,000 rps worldwide and must hold 99.995% availability with p99 ≤ 100 ms and critical durability — under $3,000/mo. The current design is single-region, under-provisioned DynamoDB, no CDN, no cache.

Scale it globally: a CDN for edge caching, an active-active multi-region compute footprint, an ElastiCache tier to shield the datastore, and DynamoDB in on-demand mode so it absorbs unpredictable global spikes without throttling or over-provisioning. Keep every tier lean — multi-region doubles the compute and datastore bill.

20,000 rps peakp99 ≤ 100ms99.995% availdurability: criticalbudget $3,000/mo

CloudFront

Networking

Enable CDN

System health

Erupting · SLA breach

/ 100

Score

SLA not met yet

Monthly cost

$4,614

Budget $3,000/mo · $1,614 over

Metrics

Capacity48

Availability50

Durability100

Cost efficiency28

Requirements

Peak capacity — 52000 rps compute · 9600 rps db (need ≥ 20000 rps)
p99 latency — ~132 ms (need ≤ 100 ms)
Availability — 99.50% (need ≥ 100.00%)
Durability — protected (need redundancy + backups)
Budget — $4614/mo (need ≤ $3000/mo)

Advisor

The database is saturated at peak — add a cache to shed read load, or scale it up.
Compute runs in a single AZ — spread across ≥2 AZs (with ≥2 instances) to meet the availability target.

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.