GitHub template

Scale runtime services independently

Jobs, schedules, realtime and telemetry run as their own services, so you scale the busy parts without scaling the rest.
Private previewRepository access is by request
Services
7+
Stores
Postgres, Redis, Fluvio, ClickHouse
Control plane
Gluon

How it runs

Boson tasks go through Redis to their own worker. Photon events go through Fluvio, and Spectra ingests them into ClickHouse.
Boson and Chronon workers take work from named pools. Add Gluon and each pool's workers can run on different hardware.

What it comes wired with

Data and work
IndraDB + Postgres/valence
Redis worker/boson
Postgres + Redis/chronon
Fluvio/photon
ClickHouse/spectra
Request context
Run and secure
Application and Docker control plane/gluon
Data services/nucleus
Secrets/secrets
Lepton
OAuth sign-in/user
Gauge
Permissions/permission
Extras
Component preview/orbital
Meson
Files
Tag
Tags/tag
Notifications
Inbox/notifications
Production isolationImplemented

Each runtime gets its own database

Your app data, jobs, schedules, live events and analytics each live in their own database. In Remote fleet, each one also runs as its own application: Valence on Postgres, Boson on Redis, Photon on Fluvio, Spectra on ClickHouse. A heavy analytics query can't slow down sign-in, and a backlog of jobs doesn't touch the database your pages read from. Each runtime has its own connection setting, so you can move any of them to its own server.
Embedded keeps the same split, with one SQLite file per runtime inside a single process.
Planned deployerPlanned

Deploy what your code needs

The planned deployer will read your source code, work out which databases and supporting services it depends on, and provision and deploy them with your app. Adding a Boson task will bring its Redis queue, and adding a Spectra store will bring ClickHouse. Today, Remote fleet deploys through its AWS scripts and Docker Compose.

Deploy on AWS (private preview)

Remote fleet is in private preview, so the repository isn't public yet. Repository access is by request.The template ships with AWS scripts. A Docker Compose file starts Postgres, Redis, Fluvio and ClickHouse, a build script compiles each service, and a start script launches the fleet.A smoke test then checks that a Chronon worker runs jobs and that the app answers on port 3000.

Compare templates

Best forBest for one or several small applications with a limited number of users.Best for large deployments: many medium to large applications, or a large number of users.
ProcessesOne binary7+ services (see topology)
DatabasesSQLitePostgres, Redis, Fluvio, ClickHouse
Database isolationOne SQLite file per runtime, in one processIts own database per runtime, each run as its own application
Admin appsValence, Boson, Chronon, Photon, Spectra, permissions, secrets, accounts, notificationsSame admin apps as Embedded, plus Nucleus
DeployOne binary on any machineAWS scripts, with Gluon running the applications and Docker containers
Automatic deployerNot includedPlanned
AvailabilityPublicPrivate preview