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: VictoriaMetrics
providerId: victoriametrics
created: '2026-05-08'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-08, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-08'
reconciled: false
tags:
- Database
- Time-Series
- Monitoring
- Open Source
- Prometheus
- Rate Limiting
- Quotas
- Throttling
description: >-
  VictoriaMetrics is self-hosted (or VictoriaMetrics Cloud as a managed
  service); there is no managed REST quota in OSS. Operational caps are set
  via flags on each component (-search.maxConcurrentRequests, -search.maxQueryDuration,
  -insert.maxQueueDuration, -maxConcurrentInserts, -maxLabelsPerTimeseries, etc.).
  Enterprise multi-tenancy adds per-tenant cardinality / ingestion / query
  rate limits enforced at vminsert and vmselect.
notes: >-
  Numeric defaults are documented per-flag; cluster operators tune them. Cloud
  customers see per-plan caps in the Cloud console. Reconciliation flagged
  false — exact per-version defaults vary across releases.
sources:
- https://docs.victoriametrics.com/Single-server-VictoriaMetrics.html#capacity-planning
- https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html
- https://docs.victoriametrics.com/PerTenantStatistic.html
responseCodes:
  throttled: 429
  busy: 503
limits:
- name: Concurrent search requests (-search.maxConcurrentRequests)
  scope: vmselect / single-node
  metric: concurrent_requests
  limit: see -search.maxConcurrentRequests flag
  notes: Default scales with CPU count; excess requests queue or 429.
- name: Query duration cap (-search.maxQueryDuration)
  scope: vmselect / single-node
  metric: ms
  limit: see -search.maxQueryDuration flag
  notes: Long-running queries are terminated.
- name: Concurrent inserts (-maxConcurrentInserts)
  scope: vminsert / single-node
  metric: concurrent_inserts
  limit: see -maxConcurrentInserts flag
  notes: Excess inserts queue or 429.
- name: Active series per tenant (Enterprise)
  scope: tenant
  metric: active_series
  limit: per-tenant Enterprise quota
  notes: Enforced via vmstorage flags in cluster Enterprise.
- name: Ingestion rate per tenant (Enterprise)
  scope: tenant
  metric: samples_per_second
  limit: per-tenant Enterprise quota
  notes: Configurable via -storage.maxIngestionRate or similar.
policies:
- name: Backoff Strategy
  description: Honor 429/503 with exponential backoff; vmagent buffers and retries on remote-write failures.
- name: Capacity Planning
  description: Right-size vminsert, vmselect and vmstorage per the cluster sizing guide.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com