BigQuery Proxy Deployment — Cloud, BYOC & On-Premises

Choose the deployment model that fits your organization. Fully managed cloud, bring your own GCP project, or run on-premises — same proxy, same savings, same features.

CL

Cloud

Fully Managed

We host and operate everything. Create an account, add your BigQuery credentials, and start proxying queries in under five minutes. No servers to provision, no infrastructure to maintain.

What you get

  • Zero infrastructure to manage
  • Automatic updates and security patches
  • Built-in monitoring, query logs, and merge analytics
  • Team management with role-based access
  • Google OAuth single sign-on
  • Multiple BigQuery projects per account
  • Dedicated proxy endpoint per organization

How it works

1 Sign up at app.queryfuser.com
2 Add your BigQuery project and upload a service account key
3 Point your BI tool at the provided proxy host and port
Get Started Free

Free during beta • No credit card required

BY

Bring Your Own Cloud

Your GCP Project

Run QueryFuser inside your own GCP project. Your queries and data never leave your infrastructure. The proxy runs on your Compute Engine VM and uses the VM’s service account to access BigQuery directly — no credentials to upload or manage.

What you get

  • Data stays entirely within your GCP project
  • No credentials to upload — uses the VM’s service account
  • One-command setup script
  • All merge features — same engine as cloud
  • Local dashboard for query logs and merge analytics
  • License-based — phones home every 4 h with usage metrics (see below)
  • Automatic project creation on first registration

How it works

1 Generate a license key from your cloud account
2 Run the setup script on a GCE VM in your project
3 Register a local account and start querying
Get a License Key

Requires a cloud account first

OP

On-Premises

Any Server

Run QueryFuser on any Linux server — your own data center, AWS, Azure, or any other environment. Authenticate to BigQuery with a service account JSON key uploaded through the dashboard. Same engine, same features, no GCP VM required.

What you get

  • Run on any Linux server with Docker
  • Full control over infrastructure and networking
  • Upload service account JSON per BigQuery project
  • All merge features — same engine as cloud and BYOC
  • Local dashboard for query logs and merge analytics
  • Works behind corporate firewalls (outbound HTTPS only)
  • License-based — phones home every 4 h with usage metrics (see below)

How it works

1 Generate a license key from your cloud account
2 Run the setup script on your server
3 Upload a service account key and start querying
Get a License Key

Requires a cloud account first

Side by Side

Feature Comparison

Feature Cloud BYOC On-Prem
Query merging & fusion ✓ ✓ ✓
PostgreSQL wire protocol ✓ ✓ ✓
Automatic SQL translation ✓ ✓ ✓
Merge logs & analytics ✓ ✓ ✓
Automatic fallback on failure ✓ ✓ ✓
Configurable merge window ✓ ✓ ✓
Infrastructure management Managed by us You manage You manage
Data residency Our GCP project Your GCP project Your servers
Credential handling Upload SA key VM service account Upload SA key
Authentication Email + Google OAuth Email / password Email / password
Updates Automatic docker compose pull docker compose pull
Network latency to BigQuery Our region → BQ Same project Your network → BQ
Runs on GCP Yes (managed) Yes (your project) No — any server
BYOC Details

What Runs in Your Project

The BYOC deployment runs the same QueryFuser binary as the cloud version. Here’s what gets deployed into your GCP project.

CE

Compute Engine VM

A single VM (e2-small or larger) runs the QueryFuser proxy. The setup script installs dependencies, pulls the binary, and configures systemd to start on boot.

PG

PostgreSQL (Docker)

A PostgreSQL 16 container runs alongside the proxy in Docker Compose. It stores user accounts, project configs, query logs, and merge analytics. Data is persisted in a Docker volume. You can optionally migrate to Cloud SQL or any external PostgreSQL for managed backups and HA.

SA

Service Account

The VM’s default service account accesses BigQuery. No JSON keys are created or stored. Grant it bigquery.dataViewer, bigquery.jobUser, bigquery.user, and optionally bigquery.metadataViewer.

License Phone-Home

Both BYOC and on-premises instances validate their license every 4 hours by sending a lightweight heartbeat to our API. The payload contains the license key and aggregate usage metrics: query count, bytes scanned, bytes saved by merging, and bytes saved by caching. No SQL queries, table names, column names, or row-level data is ever transmitted. If the license expires or is revoked, the proxy continues serving queries for a 72-hour grace period before shutting down.

Which is Right for You?

Choosing a Deployment Model

Choose Cloud if…

  • You want to start in under 5 minutes with zero infrastructure
  • Your team is comfortable uploading a service account key
  • You prefer automatic updates and managed operations

Choose BYOC if…

  • Your data must stay within your own GCP project
  • You don’t want to distribute service account keys
  • You need the lowest possible latency to BigQuery
  • Compliance requires infrastructure you control

Choose On-Prem if…

  • You’re not on GCP (AWS, Azure, on-prem servers)
  • Your security policy doesn’t allow third-party VMs in your cloud
  • You need to run behind a corporate firewall
  • You want full control over every layer of the stack

Ready to Deploy?

All three deployment models use the same proxy engine. Start with cloud and move to BYOC or on-premises later — your choice.

Get Started Contact Us