Deploy and Host Cube 1.8 on Railway

Cube is the open-source semantic layer: define metrics and dimensions once, then query them from apps, BI tools and AI agents over REST and GraphQL with consistent numbers and fast cached answers. This template runs Cube Core 1.8.22 in its production layout. Connect your own database with CUBEJS_DB_* on Cube API. Until then, a demo cube answers from DuckDB in memory, so the API works right after deploy.

About Hosting Cube 1.8

A production Cube deployment has four parts: API instances, a refresh worker that builds pre-aggregations, Cube Store (router and workers) that stores and serves them, and storage that Cube Store's nodes share. This template wires all of them on Railway's private network, with a Railway Bucket as Cube Store's S3 storage. Only the API is public. Every request needs a JWT signed with a generated secret, and dev mode (the Playground) is off. The data model lives in a variable, so you can edit it in Railway without a repository, or move it to git later.

Common Use Cases

One definition of revenue, active users or churn for dashboards, apps and AI agents
Customer-facing analytics in a SaaS app, with row-level security from the JWT
Fast APIs over large tables: pre-aggregations answer in milliseconds
A governed metrics layer for LLM agents instead of raw SQL

Dependencies for Cube 1.8 Hosting

Cube 1.8.22 (cubejs/cube), API and refresh worker
Cube Store 1.8.22 (cubejs/cubestore), router and worker
Railway Bucket (S3 storage for Cube Store)

Deployment Dependencies

Cube docs: https://docs.cube.dev
Production checklist: https://docs.cube.dev/admin/deployment/core#production-checklist
Data model reference: https://docs.cube.dev/reference/data-modeling/cube
REST API: https://docs.cube.dev/reference/core-data-apis/rest-api
Cube on GitHub: https://github.com/cube-js/cube

Implementation Details

What you set: nothing to start; your database later. The API secret and the sample model are ready.

After deploy: copy CUBEJS_API_SECRET from the Cube API variables, mint a JWT and query:

CUBEJS_API_SECRET="paste-from-Cube-API-variables"
b64url() { openssl base64 -A | tr '+/' '-_' | tr -d '='; }
H=$(printf '{"alg":"HS256","typ":"JWT"}' | b64url)
P=$(printf '{"exp":%d}' $(( $(date +%s) + 3600 )) | b64url)
S=$(printf '%s.%s' "$H" "$P" | openssl dgst -sha256 -hmac "$CUBEJS_API_SECRET" -binary | b64url)
curl -s "https://your-cube-api.up.railway.app/cubejs-api/v1/load" -H "Authorization: Bearer $H.$P.$S" \
  -G --data-urlencode 'query={"measures":["orders.count"],"dimensions":["orders.status"]}'

The answer comes from the orders_by_month pre-aggregation in Cube Store, built by the refresh worker a minute or two after the first deploy.

Data model: edit CUBE_MODEL (YAML cubes and views) on Cube API, then redeploy Cube API and Cube Refresh Worker. For bigger models, keep a model/ folder in a repo with a two-line Dockerfile (FROM cubejs/cube:v1.8.22, COPY model /cube/conf/model) and point both Cube services at it; the README shows how. CUBE_CONFIG_JS takes an optional cube.js for row-level security or multi-tenancy.

This template:

Pins Cube and Cube Store 1.8.22
Runs the production layout: API, refresh worker, Cube Store router + worker, Railway Bucket
Keeps Cube Store and the refresh worker private; only the API is public
Requires a JWT (HS256, generated 64-character secret) on every request; dev mode off
Writes the data model from a variable, since Railway can't mount files
Starts with a DuckDB demo cube, so it is healthy before you connect a database
Keeps built pre-aggregations in the bucket across redeploys
Healthchecks on Cube API, the refresh worker and the Cube Store router (/readyz)

Scaling: add Cube Store workers by duplicating the worker service and listing it in CUBESTORE_WORKERS; one worker is enough for small and medium data. The SQL API for BI tools is off by default; the README shows how to turn it on with a password.

Resources: about 0.75 GB RAM in total at idle (measured on Railway). Licence: Apache 2.0 (Cube Core).

Why Deploy Cube 1.8 on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying Cube 1.8 on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.