Cloud Architecture Simulation Benchmark

AWS vs GCP vs Azure vs OCI vs DO: Multi-Zone Burst Traffic Benchmark

10-Minute Simulation  ·  2× Burst Load  ·  Five Cloud Providers
Powered by Cloud World Model — simulate cloud architecture without the cloud bill
AWS — us-east-1 — 3× m5.large + RDS r5.large + ALB
GCP — us-central1 — 3× e2-standard-2 + Cloud SQL + CLB
Azure — East US — 3× D2s_v3 + SQL S2 + Standard LB
OCI — us-ashburn-1 — 3× E4.Flex + MySQL HeatWave + LB
DO — nyc1 — 3× s-4vcpu-8gb + Managed DB + LB
Scenario Configuration
Parameter AWS (us-east-1) GCP (us-central1) Azure (East US) OCI (us-ashburn-1) DO (nyc1)
Compute 3× EC2 m5.large (2 vCPU / 8 GB each) 3× e2-standard-2 (2 vCPU / 8 GB each) 3× Standard_D2s_v3 (2 vCPU / 8 GiB each) 3× VM.Standard.E4.Flex (1 OCPU / 8 GB each) 3× s-4vcpu-8gb Droplet (4 vCPU / 8 GB each)
Database RDS db.r5.large (2 vCPU / 16 GB, MySQL) Cloud SQL db-n1-standard-2 (2 vCPU / 7.5 GB) SQL Database Standard S2 (50 DTUs) MySQL HeatWave DB System (2 OCPU / 16 GB) Managed DB db-s-2vcpu-4gb (2 vCPU / 4 GB)
Load Balancer Application Load Balancer (ALB) Cloud Load Balancing (CLB) Standard Load Balancer OCI Load Balancer Standard Load Balancer
Traffic Pattern Ramp from 1 k RPS baseline → 2 k RPS (2× burst) over 6 steps; sustained for 4 steps
Pricing Source $0.55/hr cluster (official June 2026) $0.41/hr cluster (official June 2026) $0.41/hr cluster (official June 2026) $0.29/hr cluster (official June 2026) $0.30/hr cluster (official June 2026)
Performance & Cost
P95 Latency — 10-Minute Simulation
Annotated: burst traffic begins at minute 4, autoscale triggers at minute 6
Cost/hr at Baseline / 1.5× / 2× Burst
OCI delivers the lowest hourly cost; DigitalOcean is comparable at $0.30/hr
KPI Scorecards
Key Takeaways
Simulation Findings — 5 Quotable Conclusions
1
OCI costs 47% less than AWS and 29% less than GCP/Azure at peak ($0.29/hr vs $0.55 and $0.41/hr) — VM.Standard.E4.Flex + MySQL HeatWave deliver the lowest per-hour cluster bill across all three traffic levels, making OCI the clear cost winner in this tier.
2
DigitalOcean matches OCI on cost ($0.30/hr) but pays a latency tax — the s-4vcpu-8gb cluster peaks at 256 ms P95 vs 123 ms for OCI E4.Flex at the same price point, because the Managed DB db-s-2vcpu-4gb’s lower connection ceiling limits headroom under burst.
3
Mid-tier compute amplifies P95 tail latency under burst — e2-standard-2 peaks at 227 ms P95 and D2s_v3 at 184 ms at 2× burst, vs 119 ms for AWS m5.large, showing how per-instance headroom is the dominant factor for tail-latency control.
4
Azure D2s_v3 outperforms GCP e2-standard-2 at identical cluster cost — both clusters cost $0.41/hr, but D2s_v3’s higher per-instance throughput ceiling (5 k vs 4 k RPS) keeps P95 latency 43 ms lower at peak (184 ms vs 227 ms).
5
AWS and OCI maintain sub-125 ms P95 at 2× burst — m5.large (119 ms) and E4.Flex (123 ms) absorb the 2× load with minimal tail-latency impact, while GCP, Azure, and DigitalOcean clusters show measurably higher P95 under the same traffic ramp.
Methodology
About this report: All metrics were produced by the Cloud World Model simulation engine — a deterministic, capacity-aware model that calculates CPU utilisation, P50/P95 latency, throughput, error rates, and cost from first-principles resource profiles without provisioning any real cloud infrastructure. Official hourly rates from AWS, GCP, Azure, OCI, and DigitalOcean pricing calculators (June 2026, us-east-1 / us-central1 / East US / us-ashburn-1 / nyc1) are embedded directly in the engine and validated weekly. See cloudworldmodel.ai/fidelity for the full accuracy specification (cost ±10%, latency ±15%).

Mid-tier compute and burst headroom: The GCP and Azure configurations use e2-standard-2 and Standard_D2s_v3 respectively — 2 vCPU shapes chosen to reflect realistic mid-tier deployments (see the Capacity column in the scenario table above). Smaller per-instance headroom means these clusters reach higher CPU utilisation at 2× burst traffic, which is why their P95 latency (227 ms / 184 ms) and error rates (10–12%) are elevated compared to the larger m5.large and E4.Flex shapes. This is expected behaviour, not a measurement artefact: the numbers reflect the trade-off between instance cost and burst tolerance inherent to mid-tier compute.

Simulate your own architecture — free

Drop in your own resources, run burst traffic patterns, and compare AWS, GCP, Azure, OCI, and DigitalOcean side-by-side — no cloud account required.

Try Cloud World Model →