Deployment model

PackDB is a fully managed cloud service. There is no database for you to deploy, scale, or operate — you point your telemetry at it and query.

What you run

On your side, the only moving part is whatever already produces telemetry: your applications, an OpenTelemetry collector, or both. PackDB runs storage, indexing, query, and retention for you.

→Direct from your apps. Export OTLP straight from your OpenTelemetry SDKs to the PackDB ingest endpoint. Good for getting started and for serverless workloads where running a collector is awkward.
→Through a collector. Run an OpenTelemetry collector as a gateway or sidecar and point its exporter at PackDB. Good for batching, retries, and applying processors before data leaves your network. This is the path the quickstart uses.

Regions and isolation

Each workspace is an isolated tenant with its own API keys, resource limits, and quota enforcement, so every workspace runs against its own bounded envelope for ingest and query load. The ingest and query URLs are shared per environment; the packdb_… key carries the workspace binding. The Connect panel always shows the right environment endpoints and key flow.

Dedicated deployments

Need a dedicated cluster, a specific region, or a private networking arrangement? Those are handled as enterprise engagements rather than self-serve. Talk to us →