Azure Databricks · Rate Limits

Microsoft Azure Databricks Rate Limits

Azure Databricks REST APIs apply per-workspace and per-API rate limits. Most APIs target a few hundred requests per second per workspace; Jobs API and Files API have lower per-second targets. Some APIs (Statement Execution, Model Serving) use endpoint-specific limits. SCIM and account-level APIs are throttled separately.

Microsoft Azure Databricks Rate Limits is the machine-readable rate-limit profile for Azure Databricks on the APIs.io network, conforming to the API Commons Rate Limits specification.

It captures 5 rate-limit definitions, measuring requests_per_second, concurrent_runs, concurrent_statements, and queries_per_second.

The profile also includes 4 backoff/retry policies defined and response codes documented for throttled and serviceUnavailable.

Tagged areas include Rate Limiting, Analytics, and Microsoft Azure.

5 Limits Throttle: 429
Rate LimitingAnalyticsMicrosoft Azure

Limits

Workspace API requests workspace
requests_per_second
see Databricks limits page (typical hundreds RPS per workspace)
Jobs API runs workspace
concurrent_runs
workspace-configurable max concurrent runs
Statement Execution API workspace
concurrent_statements
see Statement Execution API limits
Model Serving endpoint QPS endpoint
queries_per_second
configurable per endpoint with concurrency budget
SCIM (account) account
requests_per_second
see SCIM provisioning limits

Policies

Honor Retry-After
When the API returns 429, wait the duration specified in Retry-After before retrying.
Backoff with jitter
Apply exponential backoff with jitter to avoid thundering herds against the workspace control plane.
Bulk endpoints over loops
Prefer batch endpoints (e.g. Jobs runs list with filters) and Delta Live Tables over polling individual resources.
Workspace partitioning
Heavy multi-tenant workloads should be partitioned across multiple workspaces to avoid control-plane throttling.

Sources