- About scaling servers
- Horizontal scaling
- Vertical scaling
About scaling servers
You can scale your servers in two ways: horizontal and vertical. Horizontal scaling involves adding more servers, whereas vertical scaling involves altering the resources of a specific server, for example increasing the server size.
Horizontal scaling works differently for each server type, and is only available if you have deployed using your cloud provider.
To scale up your web servers, start by adding a load balancer to your stack, which will distribute traffic to your servers. Next, from your stack detail page, click the link to your web server group (eg. Docker server). To add a web server, click Scale up in the top right corner, select your desired server size and quantity, and click Scale up. Your new server(s) will automatically be added to the load balancer after they have completed provisioning and deployment, ready to serve traffic.
If you are using AWS, you will also have the option to scale your servers to different Availability Zones within your region.
You can also scale down your web servers. From your stack detail page, click the link to your web server group (eg. Docker server), and click the X icon next to the server you would like to scale down. This server will automatically be removed from your load balancer, but you will need to delete it from your cloud provider. Note that your primary web server cannot be scaled down, because this would leave you without a web server.
When you first build your stack, your processes are run on your web server by default. To scale up a process server, click the link to your Process server group on your stack detail page. Next, click New process server in the top right corner, select the desired server size and quantity, and click Scale up.
Once the server is ready, you can move your processes from the web server to the process server by using the + and - buttons. The process server is very much like a web server, as it needs all the code and dependencies for your workers. By default however, it will not serve web content. If you would like the process server to serve web content, add a load balancer to your stack and access the load balancer page. This page allows you to toggle serving web content from a process server On and Off.
You can scale your Docker services horizontally across your Docker cluster by clicking the Docker cluster server group from your stack page, and using the + and - buttons to increase or decrease the number of running containers on each server.
In your service configuration, you can also specify constraints for a service across the cluster, which currently includes
max_count, or the max number of containers for a service across the cluster. This doesn’t affect deploys, but comes into account when scaling up/down on server or cluster.
services: <service_name>: constraints: max_count: 2
This only applies to Rails stacks not Docker ones
A number of cloud vendors allow you to increase and/or decrease the size of an existing server via their dashboard, allowing you to change the memory and CPU for existing servers. Vertical scaling works the same way for all server types.
Follow these steps to scale your server vertically:
- Shut down the server through your cloud dashboard
- Change the server size and start it up
- Wait 10 minutes for Cloud 66 to identify the change, and then redeploy the stack
Note that if you have a load balancer and are using AWS as your cloud vendor, this is slightly more complex. AWS load balancers use a unique identifier for each server, which is updated when you change the size of the server.
Scaling vertically on AWS with the instructions above will therefore only work with backend servers but not app servers, as these are served via the load balancer. For application servers, we recommend that you scale up a new server (with your desired size) and delete the old one.