Skip to content
← Back to Telopi
Security & GDPR

Security and GDPR: observability hosted in Germany

Sovereign cloud monitoring means you can say where your telemetry is, who can read it and when it is gone. This page answers each of those for Telopi as it runs today.

Where your data is stored

In Germany, on servers we run at Hetzner.

  • Telopi runs on a Kubernetes cluster of Hetzner servers in Nuremberg, Germany.

  • Your metrics, logs and traces are stored by Mimir, Loki and Tempo in Hetzner Object Storage, in the same location as the cluster.

  • Your Grafana’s dashboards and settings, and your accounts and organization, are stored in the same cluster.

What Amazon CloudFront sees

Your agents, your website’s visitors and your people reach the cluster through Amazon CloudFront, a content delivery network of Amazon Web Services. It does not store your telemetry, but it is on the way to it.

  • CloudFront ends TLS at its edge location and opens a new TLS connection, TLS 1.2 or later, to our cluster in Germany. At the edge, in between, it handles your requests unencrypted.

  • It caches nothing: caching is turned off for ingest, the Grafanas, the status reports and the status pages, and requests are passed on unchanged.

  • Plain HTTP is refused for ingest, the Grafanas, the status reports and the status pages, so an agent set to http:// fails instead of sending its key unencrypted. The only exception is Let’s Encrypt’s certificate validation.

  • The cluster’s firewall accepts HTTPS only from CloudFront and from our administrators’ addresses.

Kept apart from other customers

Every customer is a tenant of its own, with its own Grafana, credentials and limits.

  • Every request to the backends carries the tenant’s ID, set by our gateway from the route the request came in on, never by the caller.

  • A test run against a live cluster checks that one tenant can neither read nor write another tenant’s metrics, logs and traces.

  • Limits per tenant, on series, ingestion rate and log streams, stop one customer from slowing down the others.

  • Passwords are generated inside the cluster and never pass through git, CI or a person’s shell.

How long data is kept, and how it is deleted

Every piece of telemetry has a date by which it is gone.

  • By default, metrics are kept for 30 days, logs for 14 and traces for 7. The backends delete anything older on their own.

  • On the portal’s Data policy page you can change how long each signal is kept, within what your plan allows.

  • If you delete your Grafana, it stops taking telemetry at once. For seven days it can be restored; after that it is removed for good, with everything it stored.

  • When an organization is suspended, its people can no longer sign in and its ingest is refused at once. The telemetry it already stored ends with its retention.

Who operates it

Optiop GmbH, a German company, builds and runs Telopi.

  • Everything that runs in the cluster is deployed from git by Argo CD. Servers, firewall and CloudFront are Terraform, applied from GitHub Actions, not from a laptop.

  • Our staff sign in with a password and a code from an authenticator app, every time. You can turn on the same for your own accounts.

  • When our staff reveal a customer’s ingest or Grafana login, it is recorded in an audit trail.

Other services involved

Beside Hetzner and CloudFront, these services each do one job for Telopi.

  • Amazon SES, in AWS’s Frankfurt region, sends Telopi’s emails, such as invitations and password resets.

  • Google Workspace receives the mail sent to support@telopi.com.

  • Stripe handles subscriptions and payments.

  • Anthropic’s API, only if an owner switches on AI investigations: then your Grafana’s query results, your clusters’ answers and your context file are sent to it. It is off until you switch it on.

Questions about your data?

Book a demo and ask us. We will show you where your telemetry goes on a setup like yours.

Book a demo