Microsoft Azure · Rate Limits
Azure Rate Limits
Azure rate limits and quotas are defined per-service rather than at the platform level. Azure Resource Manager (ARM) enforces a default control-plane throttle of 12,000 read and 1,200 write requests per hour per subscription per region (with bursting), and each individual Azure service publishes its own data-plane and control-plane limits (e.g., Storage 20-40k req/s per account, Key Vault 2,000-4,000 per 10s per vault). See per-service quota pages for specifics.
4 Limits
Throttle: 429
Rate LimitingCloud ComputingInfrastructurePlatform as a Service
Limits
ARM read requests (control plane) subscription/region
12000
Token-bucket; bursts allowed. Refill rate enforces sustained throughput.
ARM write requests (control plane) subscription/region
1200
Tenant-level read requests (Microsoft Graph / Entra) tenant
see service-specific quotas
Per-service data-plane limits account/region/service
see service-specific quota pages (Storage, Cosmos DB, Key Vault, AKS, etc.)
Policies
Token-bucket throttling
Most Azure services use token-bucket throttling with a burst size and a refill rate. Returns 429 with Retry-After when the bucket is empty.
Quota request
Subscription-level quotas (cores, IPs, storage accounts, etc.) can be raised via the Azure portal Quotas blade or by support ticket.
Backoff
Honor Retry-After header. Use exponential backoff with jitter for transient 429/503 responses.
Per-service variation
Each Azure service publishes its own throttling matrix; consult the service-specific limits page (e.g. /azure/{service}/quotas, /azure/{service}/service-limits).