Superset Self-Hosted
IntermediateDashboardSelf-hosted Apache Superset: BI dashboards that never leave your server.
Published 27 September 2026
About Superset Self-Hosted
Superset Self-Hosted runs Apache Superset on your own server with Docker. The platform is several cooperating services: the web server that renders the UI and SQL Lab, a Celery worker and beat for asynchronous queries, alerts, and reports, and Redis for caching and the message queue. PostgreSQL is the metadata database where dashboards, charts, users, and query history live, while warehouses like ClickHouse or Snowflake sit behind Superset as data sources.
Self-hosting removes per-user BI pricing. The open-source edition has no feature caps on dashboards, charts, SQL Lab, or row-level security; Preset, the company founded by Superset's creator, sells the managed version per seat. Query text, dashboard definitions, and user records stay on your infrastructure.
The difference from Metabase is SQL Lab and breadth. Superset reads from dozens of database drivers, from PostgreSQL and MySQL to ClickHouse, BigQuery, and Snowflake, and analysts get a SQL editor next to the no-code chart builder.
The trade is operational weight. Superset's project doesn't support its Docker Compose files for production: they're a quick path to a running instance, and a real deployment sets its own secret key, backs up the metadata database, and plans upgrades, or follows the project's Kubernetes (Helm) installation, which it recommends even for a single host. Teams that want the lightest self-hosted BI should compare Metabase Self-Hosted first.
Key Features
- ✓Docker deployment of the web server, Celery worker, and beat
- ✓PostgreSQL as the metadata database for dashboards, charts, users, and query history
- ✓Redis for caching and the asynchronous query queue
- ✓Dozens of data drivers: PostgreSQL, MySQL, ClickHouse, BigQuery, Snowflake, and more
- ✓SQL Lab editor for analysts next to the no-code chart builder
- ✓Row-level security and granular roles in the open-source edition
- ✓No per-user or per-dashboard cost at any team size
When to Use Superset Self-Hosted
- →Analytics teams that work in SQL and want dashboards on the same data
- →BI over a warehouse such as ClickHouse, BigQuery, or Snowflake without per-seat licensing
- →Organizations that must keep query text and dashboards on their own infrastructure
- →Embedding dashboards in internal tools with row-level security per audience
- →Replacing a per-seat BI tool across a large team
Pros
- Zero license cost, with no feature caps on core BI
- One of the widest data-source lists in open-source BI
- SQL Lab gives analysts direct query access next to the dashboard builder
Cons
- Heavier than single-container BI tools: several services, 4 GB of memory to start
- The project doesn't support its Docker Compose files for production; hardening is yours
- Admin setup for roles, alerts, and SSO takes more initial work than Metabase
Database Options for Superset Self-Hosted
Superset's most-deployed metadata database and the Compose default: dashboards, charts, users, saved queries, and query history all land here. The default Compose volume isn't backed up, so a real deployment schedules its own dumps.
Supported for the metadata database. Choose it when your platform standard is MySQL and you'd rather not run a second engine for the BI layer, and accept that the official guides and the Compose default assume PostgreSQL, so most setup help reads that way.
Works for the metadata store wherever MySQL does, through the same driver and schema: a drop-in for shops already standardized on MariaDB. The caveat matches MySQL's, since the deployment guides assume PostgreSQL, so MariaDB earns its place through your existing estate.
These are highlighted picks. To see all the tools, check the Databases category.
Hosting Options for Superset Self-Hosted
The default: a flat-rate root VPS whose entry instance has 4 GB of memory for around €4-5/mo, Superset's starting size, and 8 GB instances, the size community admins report for a moderately used instance, cost around €8-10/mo. PostgreSQL and Redis fit on the same server.
A root Droplet with snapshots and a cloud firewall in the panel: about $24/mo for 4 GB or $48/mo for 8 GB. Managed PostgreSQL (from about $15/mo) can hold the metadata database with automated backups, which Superset's own docs say the default Compose volume doesn't have.
The budget VPS in the list: the entry KVM plan (about $6.50/mo on a two-year term) has 4 GB, and the next plan 8 GB for about $9/mo, both on two-year terms with higher renewals. The cheapest way to reach the moderate-instance size, with fewer regions than the larger providers.
An EC2 instance at roughly $30-70/mo for 4-8 GB. It makes sense once dashboards sit inside a VPC next to the warehouse they query, such as Redshift or RDS; RDS can hold the metadata database, and EKS runs the Helm chart the project recommends for production.
A Compute Engine instance at roughly $30-70/mo for 4-8 GB, worth it when Superset queries BigQuery over private paths. Cloud SQL can hold the metadata database, and GKE, free for one zonal cluster's management, runs the Helm chart the project recommends for production.
These are highlighted picks. To see all the tools, check the Hosting & Cloud category.
Reverse Proxy Options for Superset Self-Hosted
Terminates TLS in front of Superset's web server on port 8088 and routes by Docker label, so Compose restarts need no manual reloads. Set ENABLE_PROXY_FIX in superset_config.py so Superset trusts the forwarded scheme and builds HTTPS links.
The same TLS termination with a short config: automatic Let's Encrypt certificates for port 8088 and nothing to renew by hand. Caddy applies no short response timeout by default, so long SQL Lab queries aren't cut off at the proxy.
Self-Hosted PaaS Options for Superset Self-Hosted
Free and self-hosted, deploying Superset's image with PostgreSQL and Redis beside it and TLS through Coolify's bundled Traefik. Set SECRET_KEY in the environment before first start, since Superset refuses to run in production with the default key.
Superset Self-Hosted Add-ons
Each addition below extends this stack with a capability the base stack works fine without. None are required: include the ones your product actually needs when building this stack, and skip the rest.
Tunnel Add-ons
Add a tunnel when you're self-hosting without a static IP or can't open inbound ports — a home server, a VPS behind restrictive network policies, or anywhere a reverse proxy alone can't reach the internet.
An outbound-only tunnel fits an internal dashboard on a home server or behind a restrictive firewall: Superset stays reachable with no inbound ports and no static IP.
Frequently Asked Questions about Superset Self-Hosted
Why self-host Superset instead of using Preset?
Preset, the company founded by Superset's creator, sells managed Superset: free for up to 5 users, then $20 per user a month billed annually on Professional, with SSO and audit logs on Enterprise. Self-hosting replaces the per-seat bill with a fixed server and keeps dashboard definitions, query text, and user records on your infrastructure, with no feature caps in the open-source edition. The trade is running the services, backups, and upgrades yourself.
What does the server need to run?
Superset publishes no hard minimum. Its own FAQ relays community admins finding 8 GB of memory and 2 vCPUs adequate for a moderately sized instance, and the web server, worker, beat, Redis, and PostgreSQL together want 4 GB to start. 20-40 GB of SSD covers the metadata database and logs. Concurrent queries, not viewers, set the real limit, so size up when many long queries run at once.
Superset Self-Hosted or Metabase Self-Hosted?
Both are self-hosted open-source BI. Metabase is the lighter operation, one container plus a database, and the friendlier UI for non-analysts. Superset trades that for SQL Lab, a wider data-source list, and row-level security in the free edition. Pick Metabase for self-service dashboards and Superset when analysts live in SQL.
Is the Docker Compose setup ready for production?
Not out of the box, by the project's own account: Superset's docs say they don't support the Compose files for production and recommend the Kubernetes (Helm) installation, even on a single host through minikube. If you run Compose anyway, set your own SECRET_KEY, pin a released image with the image-tag Compose file, and back up the metadata database, whose default volume isn't backed up. Upgrades run database migrations with superset db upgrade, so back up before each one.
How many dashboard viewers can one server take, and why Redis?
Viewers are cheap; concurrent queries are the real limit. Redis is what keeps it that way: it caches query results so repeated dashboards don't re-run their SQL, and it's the Celery broker behind alerts, reports, and asynchronous queries. A small instance serves a team's daily dashboards comfortably; when many long SQL queries overlap, scale the web server and Celery workers first, since the metadata database is rarely what runs out.
Stacks Related to Superset Self-Hosted
Airflow Self-Hosted
InfrastructureSelf-hosted Apache Airflow: the standard data-pipeline scheduler on your infrastructure.
GitLab Self-Hosted
InfrastructureSelf-hosted GitLab with Docker: your code, your CI, your infrastructure.
Sentry Self-Hosted
InfrastructureSelf-hosted Sentry error tracking on your own infrastructure, with PostgreSQL, Redis, ClickHouse, and Kafka under the hood.
PostHog Self-Hosted
InfrastructureSelf-hosted PostHog: product analytics, session replay, and feature flags with every event on your own server.
Scores
Popularity3/5
A well-known Apache project among data teams, though less familiar than Metabase to developers outside analytics.
Learning Curve4/5
Dashboards are approachable, but running Superset means several services, a hardened config, metadata backups, and a production path the project steers toward Kubernetes.
Flexibility5/5
Dozens of database drivers, SQL Lab, a no-code chart builder, row-level security, and embedding cover nearly any BI need in the free edition.
Performance3/5
Redis caching keeps repeated dashboards fast; long concurrent queries are the limit, and the web server and workers scale out when they arrive.
Portability4/5
Dashboards export and import between instances and Preset, and Superset queries your databases in place, so no data is locked inside it.
Tools in the Superset Self-Hosted Stack
Databases
DevOps & CI/CD
BI & Analytics
Add-ons (optional — add any, or none)
Tunnel
Superset Self-Hosted Pricing
Superset, PostgreSQL, Redis, and Docker are all free; the recurring cost is the server, 4 GB to start and 8 GB for moderate use, at $6-50/mo on a VPS. Preset, the managed alternative, is free for up to 5 users and $20 per user a month beyond that, so self-hosting pays off as the team grows.
4 GB of RAM to start and 8 GB for a moderately sized instance, with PostgreSQL and Redis on the same server.
Apache-licensed with no feature caps; paid plans exist only as Preset's managed hosting.
Traefik, Caddy, NGINX, Coolify, and Dokploy all self-host free, with free Let's Encrypt certificates.
Superset Self-Hosted System Requirements
- CPU
- 2 vCPU
- RAM
- 4 GB to start, 8 GB for a moderately-sized instance
- Disk
- 20-40 GB SSD
- OS
- Any Linux with Docker
No official requirements published — tekyous guidance based on the bundle's services.
Superset's own FAQ relays community admins finding 8 GB and 2 vCPU adequate for a moderately-sized instance, and the five cooperating services (web, worker, beat, Redis, PostgreSQL) want 4 GB to start. Concurrent queries, not viewers, set the real limit.