Scaling overview
Scaling an application can mean giving one server more capacity, running the application on more than one server, or distributing traffic across those servers.
Coolify can deploy the same application image to multiple connected servers. A cloud load balancer can then direct requests to the healthy application instances. Coolify does not create or manage that external load balancer for you.
Add infrastructure when load, availability requirements, or measured bottlenecks justify it. A single appropriately sized server is easier to operate than a multi-server deployment.
Choose a scaling path
| Need | Approach | Start here |
|---|---|---|
| More CPU, memory, or storage for the current workload | Resize the current server or move the resource to a larger server | Servers overview |
| Run the same application on several standalone servers | Attach additional deployment servers to the application | Multi-server Deployments |
| Distribute public traffic and remove unhealthy servers from rotation | Put a provider-managed load balancer in front of the application servers | Cloud Load Balancing |
| Maintain an existing Docker Swarm deployment on Coolify v4 | Continue only as a temporary legacy setup | Docker Swarm (Deprecated) |
Multi-server deployment and load balancing solve different parts of the same architecture. Coolify deploys a consistent application image to each attached server. The load balancer decides which server receives each incoming request.
What Coolify handles
For a supported multi-server application, Coolify:
- keeps one server as the primary deployment server
- builds the application on the primary server or a configured build server
- pushes the built image to the Docker registry configured for the application
- deploys that image to the additional servers attached to the application
- lets you deploy, stop, remove, or promote individual servers from the application's Servers page
You remain responsible for:
- configuring the external load balancer and its health checks
- securing traffic between the load balancer and application servers
- keeping sessions, uploads, and other application state available to every instance
- operating shared databases, caches, queues, and file storage
- monitoring capacity and deciding when to add or remove servers
Prepare the application
Horizontal scaling works best when any application instance can serve any request. Before adding servers:
- Move session state, user uploads, queues, and caches out of the container filesystem.
- Use a database and storage location reachable from every application server.
- Add a health endpoint that returns a successful response only when the instance can serve traffic.
- Confirm that the application can run from the same image and configuration on every server.
- Decide where TLS terminates and how the load balancer reaches each server.
Applications with persistent storage configured in Coolify cannot use the current multi-server deployment feature. Docker Compose applications also cannot attach additional deployment servers through the application's Servers page.
Related scaling guides
This section covers infrastructure-level scaling across servers. For a provider-specific example, read Load Balancing on Hetzner. To use multiple CPU cores for a Node.js process on one server, read Node.js Multi-core Scaling.