Best practices
This section contains topics that describe some best practices for administering your system.
Best practices for distributing services
Each service has an associated service unit cost. These costs indicate how computationally expensive one service is to run compared to another. You can use these costs as a guide for how to distribute services across the instances in your system.
Best practices
- Avoid running multiple services with high service unit costs together on the same instance.
- On master instances, avoid running any services besides those classified as System services.
- To use your instances evenly, try to deploy a comparable number of service units on each instance.
Best practices for maintaining system availability
Run master instances on separate physical hardware
For a multi-instance system, master instances should run on separate physical hardware. If your instances run on virtual machines, run the master instances on separate physical hosts.
Run services on more than one instance
In a multi-instance system, you can choose which and how many instances that each service can run on. For redundancy, you should run each service on more than one instance.