specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Oracle Cloud Infrastructure
providerId: oracle-cloud
created: '2026-05-04'
modified: '2026-08-29'
generated: '2026-08-29'
method: searched
source: https://docs.oracle.com/en-us/iaas/Content/API/Concepts/usingapi.htm
sources:
  - {url: 'https://docs.oracle.com/en-us/iaas/Content/API/Concepts/usingapi.htm', http_status: 200, probed: '2026-08-29'}
  - {url: 'https://docs.oracle.com/en-us/iaas/Content/API/References/apierrors.htm', http_status: 200, probed: '2026-08-29'}
  - {url: 'https://docs.oracle.com/en-us/iaas/Content/General/Concepts/servicelimits.htm', http_status: 200, probed: '2026-08-29'}
reconciled: true
tags:
  - Cloud Computing
  - Rate Limiting
  - Oracle
description: >-
  Oracle throttles the OCI APIs but publishes no numeric rate. The only documented runtime signal is
  HTTP 429 with the body {"code":"TooManyRequests","message":"User-rate limit exceeded."}, and the
  only published remediation is client-side exponential back-off. Resource quotas — how many
  instances, VCNs, buckets or databases a tenancy may hold — are documented separately and in
  detail, but they are limits on what you can create, not on how fast you may call.
limit_count: 0
limit_count_note: >-
  Zero is the honest number. Oracle publishes no requests-per-second, per-minute or per-hour figure
  for any OCI API, in the API concepts pages, the error reference or the service-limits table. The
  429 exists; the budget is undocumented.
limits: []
response_signaling:
  status_on_exhaustion: 429
  error_code: TooManyRequests
  body: '{"code": "TooManyRequests", "message": "User-rate limit exceeded."}'
  headers:
    ratelimit_standard: false
    x_ratelimit: false
    retry_after: false
  headers_note: >-
    No RateLimit-Limit / RateLimit-Remaining / RateLimit-Reset, no X-RateLimit-*, and no Retry-After
    header is documented or declared anywhere in the nine first-party contracts in
    openapi/_original/. A client cannot see remaining budget and cannot be told when to come back.
    This is the largest runtime-semantics gap in an otherwise precisely specified API.
  correlation_header: opc-request-id
guidance:
  strategy: exponential back-off
  verbatim: >-
    "Oracle recommends implementing exponential back-off, starting from a few seconds to a maximum
    of 60 seconds."
  source: https://docs.oracle.com/en-us/iaas/Content/API/Concepts/usingapi.htm
scope:
  documented: per-user / per-tenancy, per-service, per-region
  note: >-
    Throttling is applied at the service endpoint. Oracle describes the 429 message as a "User-rate
    limit exceeded", i.e. scoped to the calling principal, but publishes no per-scope numbers.
resource_quotas:
  documented: true
  url: https://docs.oracle.com/en-us/iaas/Content/General/Concepts/servicelimits.htm
  distinction: >-
    These are capacity limits, not rate limits — e.g. 50 API Gateways per region on Universal
    Credits vs 5 on Pay As You Go; 100 TB aggregate block storage vs 30 TB; 50 load balancers vs 3.
    They vary by subscription type (Universal Credits / Pay As You Go / Trial / Always Free), by
    region and by availability domain.
  errors_when_exceeded:
    - {status: 400, code: LimitExceeded, meaning: Oracle-defined tenancy limit for the resource type.}
    - {status: 400, code: QuotaExceeded, meaning: Administrator-defined compartment quota.}
  increase_process: >-
    Service-limit increase request from the OCI Console, or via Oracle Support. See
    https://docs.oracle.com/en-us/iaas/Content/General/service-limits/create-request.htm
notes: >-
  Upgraded from method:generated to method:searched on 2026-08-29. The previous revision asserted
  "clients receive HTTP 429 with Retry-After" — the 429 is correct and documented, the Retry-After
  is not: no such header is published. Corrected here.