01 / Collect
Keep existing tracking
Receive GA4-compatible or Matomo protocol traffic, or use the native d8a web tracker.
MIT licensed Go binary Docker ready
Self-hosted analytics that stays inside your infrastructure.
An open-source analytics pipeline for teams that need control, not another black-box dashboard. Run d8a in your cloud, on-premises, or in an isolated network and write clean event data directly to storage you own.
GA4 compatible Matomo compatible No event fees
d8a / quickstart
Start the documented Docker stack
After saving the config and Compose file from the quickstart.
$ docker compose up -d 8080collector
Goruntime
Youown the data
Your own cloud
On-premises
Air-gapped mode
Auditable source
Warehouse native by design
Collection and session logic without the vendor lock-in.
d8a accepts familiar tracking protocols, computes sessions on the backend, and writes a flat schema into the destination you choose.
02 / Process
Compute sessions server-side
Session state and analytics-ready fields are handled in the backend, not reconstructed with browser-side hacks.
03 / Own
Query your own data
Use normal SQL and the BI tooling your team already trusts. Your event history is not trapped behind an API.
Bring your own storage
One pipeline. Your destination.
Write directly to a database or produce CSV and Parquet files for local or object storage. Change the reporting layer without replatforming collection.
Explore warehouse options
BigQuery 
ClickHouse 
GCS 
Local disk
For controlled environments
Put the data boundary where your policy requires it.
Deploy inside infrastructure configured for your healthcare, public-sector, or internal security requirements. d8a includes an air-gapped mode that disables telemetry and remote data refreshes by default.
- Keep collection and storage in your VPC or network
- Use local ClickHouse, filesystem, or MinIO in isolated deployments
- Audit and adapt the MIT-licensed source
d8a can support deployments designed around HIPAA or FedRAMP requirements. Compliance and authorization depend on your complete implementation, controls, and operating environment.
Scale without mystery
Simple components and documented limits.
Run the receiver and worker together for a straightforward deployment, or separate them for production. Multiple receivers are supported for ingestion availability; session state currently uses a single worker.
- Receiver
- Scale out
- Worker
- Scale up
- Published sizing guidance
- The operations guide documents approximately 7,000 requests per second on 16 cores and 30 GB RAM.
See the output
A reporting-ready schema, not a proprietary interface.
Open the live Looker Studio report backed by d8a data. It is one example: connect the same schema to Superset, Metabase, Grafana, or your team's SQL workflow.
events.sql flat schema
SELECT
session_id,
page_location,
event_name,
utm_source,
timestamp_utc
FROM events
WHERE event_name = 'page_view';
-- normal SQL, your normal tools
Open-source scope
What you get, stated plainly.
The self-hosted project is the headless analytics pipeline.
The MIT-licensed repository includes collection, protocol handling, sessionization, schema processing, and warehouse destinations. Bring the BI or dashboard layer you prefer.
The hosted service adds managed implementations.
d8a Cloud adds a managed database hosted in the EU, a built-in reporting layer, and Ask AI for exploring your analytics. The self-hosted pipeline has no dependency on these cloud services.
Contributions are reviewed by humans.
Bug fixes and focused improvements are welcome. Discuss larger architecture changes with maintainers on Discord or GitHub Issues first.
Your infrastructure. Your data.
Read the code before you trust the pipeline.
Start with the repository, run the documented Docker setup, and ask the maintainers anything the code does not answer.