Provider Coverage

Explore the complete service, size-catalog and billing-model coverage for AWS, GCP, Azure, OCI and DigitalOcean. Service-level labels distinguish direct benchmark validation from behavior derived by the model.

How to read this coverage reference

Coverage is sourced from the simulator's shared provider catalog and pricing constants. Size counts describe catalog entries, not every available SKU from a cloud provider. Base latency and compute rate are reference values used by the model, not a measured latency promise or a complete workload bill. “Measured” means the service is directly exercised in the accuracy benchmark; “Extrapolated” means behavior is derived from the simulation model without a direct benchmark validation pass.

Coverage snapshot reviewed: 2026-10-31. Use the service names and labels below to check fit for a proposed architecture, then read the linked accuracy and fidelity evidence for assumptions and limits.

AWS service coverage

AWS models 1157 compute sizes, 688 managed database sizes and 5 storage sizes. Reference base compute rate: $0.096 per hour; intra-region base latency: 45 ms. These values are catalog-derived reference inputs, not a quote for a complete architecture.

Compute

  • EC2 (t2, t3, m5, m6i, m7a, m7g, m7gd, m7i, m7i-flex, m8a, m8azn, m8g, m8gb, m8gd, m8gn, m8i, m8i-flex, m8ib, m8id, m8idb, m8idn, m8in, m8ine, m9g, m9gd, c5, c6i, r5, r6i) — Measured
  • ec2: expanded provider-published instance catalog (915 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated

Network

  • Application Load Balancer (ALB) — Measured
  • CloudFront CDN — Extrapolated
  • NAT Gateway — billing: hourly + GB — Extrapolated
  • VPC Endpoint (Interface) — billing: hourly — Extrapolated
  • Elastic IP (unattached) — billing: hourly — Extrapolated

Database

  • RDS (PostgreSQL / MySQL, Graviton3 families) — Measured
  • RDS SQL Server Express (T3 Unlimited) — billing: Instance hours + storage + caller-defined CPU credits (us-east-1, Single-AZ) — Extrapolated
  • Aurora PostgreSQL — Measured
  • Aurora DSQL — billing: request-based — Extrapolated
  • Aurora Serverless v2 — billing: ACU-hour — Extrapolated
  • DynamoDB — billing: request-based — Extrapolated
  • aurora-postgresql: expanded provider-published instance catalog (91 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated
  • rds-postgresql: expanded provider-published instance catalog (296 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated
  • rds: expanded provider-published instance catalog (285 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated

Cache

  • ElastiCache (Redis) — Measured
  • elasticache: expanded provider-published instance catalog (73 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated

Storage

  • S3 — billing: GB-month — Extrapolated
  • EBS (sc1, st1, gp3, io1, io2 Block Express) — Measured

Serverless

  • Lambda — billing: GB-second — Extrapolated
  • ECS Fargate (Linux task fleets) — billing: vCPU-second + GB-second — Extrapolated
  • App Runner — billing: vCPU-second — Extrapolated

Observability

  • CloudWatch — billing: GB-ingested — Extrapolated

Kubernetes

  • Amazon EKS — Extrapolated
  • ROSA Hosted Control Planes (AWS) — published $0.171/4-vCPU worker service fee + $0.25 cluster fee — billing: platform + control plane + AWS workers — Extrapolated
  • ROSA Classic (AWS) — published $0.171/4-vCPU worker service fee — billing: platform + AWS workers — Extrapolated
  • OpenShift Dedicated (AWS) — OpenShift platform overlay; subscription pricing estimated — billing: platform fee + AWS workers — Extrapolated
  • Self-managed OpenShift (AWS) — subscription pricing estimated — billing: AWS workers + estimated subscription — Extrapolated
  • EKS GPU node group (g5 / NVIDIA A10G) — Measured
  • EKS large-GPU node group (p4d.24xlarge 8×A100, p3.16xlarge 8×V100, g5.48xlarge 8×A10G) — Extrapolated

Queue / Streaming

  • Amazon SQS — billing: request-based — Extrapolated

GCP service coverage

GCP models 377 compute sizes, 29 managed database sizes and 4 storage sizes. Reference base compute rate: $0.100 per hour; intra-region base latency: 48 ms. These values are catalog-derived reference inputs, not a quote for a complete architecture.

Compute

  • Compute Engine (e2, n2, n2d, c2) — Measured
  • compute: expanded provider-published instance catalog (321 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated

Network

  • Cloud Load Balancing — Measured
  • Cloud CDN — Extrapolated
  • Cloud Armor — billing: request-based — Extrapolated
  • Cloud NAT — billing: hourly + GB — Extrapolated
  • Private Service Connect Endpoint — billing: hourly — Extrapolated
  • Static External IP — billing: hourly — Extrapolated

Database

  • Cloud SQL (Shared Core, N1, N2, Enterprise, Enterprise Plus) — Measured
  • Cloud Spanner — Extrapolated
  • Bigtable (SSD / HDD) — Extrapolated
  • Firestore — billing: request-based — Extrapolated

Cache

  • Memorystore (Redis) — Measured

Storage

  • Cloud Storage — billing: GB-month — Extrapolated
  • Persistent Disk (pd-standard, pd-balanced, pd-ssd, pd-extreme) — Measured

Serverless

  • Cloud Run — billing: vCPU-second — Extrapolated
  • Cloud Functions — billing: request + vCPU/GiB-second — Extrapolated
  • App Engine (Standard F2) — billing: instance-hour — Extrapolated

Observability

  • Cloud Operations Suite (Logging + Monitoring + Trace) — billing: GB-ingested — Extrapolated

Kubernetes

  • GKE Autopilot — Extrapolated
  • OpenShift Dedicated (GCP) — OpenShift platform overlay; subscription pricing estimated — billing: platform fee + GCP workers — Extrapolated
  • Self-managed OpenShift (GCP) — subscription pricing estimated — billing: GCP workers + estimated subscription — Extrapolated
  • GKE GPU node pool (NVIDIA T4) — Measured
  • GKE large-GPU node pool (a2-highgpu-4g/8g, a2-megagpu-16g — NVIDIA A100) — Extrapolated
  • NVIDIA DGX Cloud B200 on GCP (dgx-b200, 8× B200 SXM) — DGX Cloud managed AI platform — Extrapolated

Queue / Streaming

  • Cloud Pub/Sub — billing: request-based — Extrapolated

Azure service coverage

Azure models 251 compute sizes, 19 managed database sizes and 8 storage sizes. Reference base compute rate: $0.096 per hour; intra-region base latency: 50 ms. These values are catalog-derived reference inputs, not a quote for a complete architecture.

Compute

  • Virtual Machines (B burstable, Dv3, Dv4, Fv2 compute-optimized) — Measured
  • App Service (Free F1, Basic B1, Standard S1, Premium P1v3) — Extrapolated
  • vm: expanded provider-published instance catalog (213 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated

Network

  • Application Gateway / Load Balancer — Measured
  • Azure CDN — Extrapolated
  • Front Door (Standard) — billing: request + GB — Extrapolated
  • Azure NAT Gateway — billing: hourly + GB — Extrapolated
  • Azure Private Endpoint — billing: hourly — Extrapolated
  • Azure Static Public IP — billing: hourly — Extrapolated

Database

  • Azure SQL (Basic, Standard DTU, vCore General Purpose, Hyperscale, Business Critical) — Measured
  • SQL Managed Instance (General Purpose, Business Critical) — Measured
  • Cosmos DB — Extrapolated
  • PostgreSQL Flexible Server (B1ms, D4s_v3) — Extrapolated

Cache

  • Cache for Redis — Measured

Storage

  • Blob Storage — billing: GB-month — Extrapolated
  • Managed Disks (Standard HDD/SSD, Premium SSD, Ultra Disk) — Measured

Serverless

  • Azure Functions — billing: request-based — Extrapolated
  • Azure Container Apps — billing: request + vCPU/GiB-second — Extrapolated

Observability

  • Azure Monitor (Log Analytics) — billing: GB-ingested — Extrapolated

Security

  • Defender for Cloud (Server Plan 2) — billing: server-month — Extrapolated

Kubernetes

  • Azure AKS — Extrapolated
  • Azure Red Hat OpenShift (ARO) — published East US $0.171/4-vCPU license line — billing: platform + Azure workers — Extrapolated
  • Self-managed OpenShift (Azure) — subscription pricing estimated — billing: Azure workers + estimated subscription — Extrapolated
  • AKS GPU node pool (NC A10 v4-series) — Measured
  • AKS large-GPU node pool (ND96amsr_v4, ND96asr_v4, NC96ads_A100_v4 — NVIDIA A100) — Extrapolated
  • NVIDIA DGX Cloud H200 on Azure (dgx-h200, 8× H200 SXM5) — DGX Cloud managed AI platform — Extrapolated

Queue / Streaming

  • Azure Service Bus — billing: request-based — Extrapolated

OCI service coverage

OCI models 42 compute sizes, 15 managed database sizes and 6 storage sizes. Reference base compute rate: $0.049 per hour; intra-region base latency: 52 ms. These values are catalog-derived reference inputs, not a quote for a complete architecture.

Compute

  • Compute Flex (E3 Flex, E4 Flex, Standard3 Flex, Optimized3 Flex) — Measured
  • Bare Metal (Standard E4, Standard3, Optimized3) — Extrapolated
  • VM.Standard.E2.1.Micro (Always Free) — Extrapolated
  • OCI Container Instances — billing: OCPU-hour — Extrapolated
  • oci-compute: expanded provider-published instance catalog (8 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated

Network

  • Flexible Load Balancer — Measured
  • OCI CDN — billing: GB transferred — Extrapolated
  • OCI WAF — billing: request-based — Extrapolated
  • OCI DNS — billing: query-based — Extrapolated
  • OCI Reserved Public IP — billing: hourly — Extrapolated

Database

  • MySQL HeatWave (E4 Flex, 1–32 OCPU) — Measured
  • Autonomous Database (8/16 OCPU) — Measured
  • Autonomous DB (Always Free) — Extrapolated
  • Exadata X8M (2-node) — Extrapolated

Storage

  • Block Volume (Balanced, Higher Performance, Ultra High Performance) — Measured
  • OCI Object Storage — billing: GB-month — Extrapolated
  • OCI File Storage — billing: GB-month — Extrapolated

Serverless

  • OCI Functions — billing: GB-second — Extrapolated

Observability

  • OCI Logging — billing: GB-ingested — Extrapolated

Cache

  • OCI Cache with Redis — Extrapolated

Kubernetes

  • OCI Kubernetes Engine (OKE) — Extrapolated
  • OKE GPU node pool (VM.GPU.A10.1) — Measured
  • NVIDIA GB200 NVL72 on OCI (gb200-nvl72, 72× Blackwell) — DGX Cloud managed AI platform — Extrapolated

Queue / Streaming

  • OCI Streaming — billing: request-based — Extrapolated

DigitalOcean service coverage

DigitalOcean models 39 compute sizes, 8 managed database sizes and 1 storage sizes. Reference base compute rate: $0.033 per hour; intra-region base latency: 55 ms. These values are catalog-derived reference inputs, not a quote for a complete architecture.

Compute

  • Droplets — Basic (1–32 vCPU) — Measured
  • Droplets — Premium NVMe (Intel & AMD, 2–32 vCPU) — Measured
  • Droplets — CPU-Optimized (c-4) — Extrapolated
  • droplet: expanded provider-published instance catalog (9 added sizes) — billing: Reference configuration only; price/hardware sourced, performance estimated — Extrapolated

Network

  • Load Balancer (Standard) — Measured
  • Spaces CDN — billing: GB egress — Extrapolated
  • DigitalOcean Reserved IP — billing: hourly — Extrapolated

Database

  • Managed Databases — Basic (PostgreSQL / MySQL) — Measured
  • Managed Databases — General Purpose — Measured
  • Managed Databases — Memory-Optimized — Extrapolated
  • Managed MongoDB — Extrapolated

Cache

  • Managed Redis — Measured

Queue / Streaming

  • Managed Kafka — Extrapolated

Kubernetes

  • DOKS (Managed Kubernetes) — Measured
  • DOKS GPU worker pool (H100 GPU Droplet) — Measured

Storage

  • Spaces (Object Storage) — billing: GB-month — Extrapolated
  • DigitalOcean Volume Snapshot — billing: GB-month — Extrapolated

Observability

  • DO Monitoring — billing: free — Extrapolated

Serverless

  • App Platform (Basic + Pro) — Extrapolated

Review the current accuracy methodology and results · Review model fidelity and assumptions · Read the multi-cloud burst traffic simulation report