Skip to main content
Some aiAxonIQ deployments run in the customer’s own environment rather than as a hosted service. If yours is one, this page is what changes for you.
Self-hosting is not self-service today. There is no public download, no published installer and no “run this to stand up aiAxonIQ” guide in this documentation. A self-hosted deployment is arranged with your account contact, and the operational material — installation, upgrade, backup, restore and monitoring of the platform itself — is delivered with the deployment, versioned against the exact release you are running.That is deliberate rather than an omission. Operational procedures that drift from the release they describe are worse than no procedures, and a public page cannot be pinned to your version.

What is different for you

Almost nothing about using the product. Everything in this documentation applies — the same API, the same query syntax, the same instrumentation, the same guides. Three things change:
1

Your endpoint is your own domain

Wherever these pages write $OIQ_ENDPOINT, yours is on your domain rather than a hosted one. Get Started in your dashboard shows the correct value for your deployment — it is the authority, because no static page can know it.See Ingest endpoints.
2

Your operator controls plan features and limits

Feature entitlements, retention and volume allowances are set by whoever runs your deployment rather than by a subscription. A 403 UPGRADE_REQUIRED is a conversation with them, not with billing. See Plan limits.
3

Some capabilities depend on your configuration

Email delivery, single sign-on and outbound webhook notifications are each configured per deployment. If one is not working, it is usually not configured rather than broken — see Below.

What runs

A self-hosted deployment is a container stack: the ingest receiver, the API, the dashboard, the processing workers, the AI service, and the data stores behind them — an analytical store for telemetry, a search index for log full-text, a relational control plane, a cache and a vector store. The architecture is identical to the hosted service, which is why every page in Concepts applies to you unchanged.
Kubernetes manifests exist but are not the deployed shape. Deployments today are container-orchestrated on hosts you provide. If a cluster deployment is a requirement for you, raise it with your account contact rather than assuming from the presence of manifests.

Things that depend on your deployment

All of these need an SMTP provider configured. Without one they fail with an explicit error rather than appearing to succeed, which is the correct behaviour and is still confusing if you do not know to look.If invitations are not arriving, this is the first thing to check with your operator.
SAML is configured per deployment. The status endpoint reports whether it is enabled, which is how a sign-in page decides whether to offer it. See Single sign-on.
The generic outbound webhook channel is disabled by default and gated behind a flag your operator controls. Slack, PagerDuty, Teams, OpsGenie and email are unaffected. See Alerting.
Reachable only where your reverse proxy has been configured to carry it. OTLP over HTTP works everywhere and is the better default in most networks anyway. See Exporters.
Scheduled backups of the control plane and telemetry stores are part of the deployment, and the procedure for verifying and restoring them ships with it, matched to your release.Ask your operator when a restore was last rehearsed. A backup that has never been restored is a hypothesis. This is the single most valuable question on this page.

Getting telemetry in

Identical to every other deployment, with your own endpoint:

Installation overview

Collector or direct, and which platform guide to open.

Docker

Containers on one or a few hosts.

Kubernetes

A cluster, a DaemonSet and a gateway.

Questions for your operator

Worth asking early, because each has a different answer per deployment and each changes what you should do:
  1. What is my ingest endpoint, and does it carry gRPC as well as HTTP?
  2. Is email configured? Invitations, resets and one-time codes all depend on it.
  3. Is single sign-on configured?
  4. What is my retention window, and what are my volume allowances?
  5. When was a restore last rehearsed?
  6. How do I request an upgrade, and what is the current version?

Next

Platform architecture

What runs, and why there are four stores.

Ingest endpoints

Paths, status codes and limits.

Enterprise onboarding

Rolling out across teams.