The best practice is to totally remove any unneeded services and utilities on a system to prevent any chance of compromise or use. If they are just disabled, it is possible for them to be inadvertently started again at any point, or another exploit could be used to start them again.
Removing also negates the need to patch and maintain them going forward.

Explanation/Reference:
Explanation:
Fault tolerance allows a system to continue functioning, even with degraded performance, if portions of it fail or degrade, without the entire system or service being taken down. It can detect problems within a service and invoke compensating systems or functions to keep functionality going. Although redundancy is similar to fault tolerance, it is more focused on having additional copies of systems available, either active or passive, that can take up services if one system goes down. Elasticity pertains to the ability of a system to resize to meet demands, but it is not focused on system failures. Automation, and its role in maintaining large systems with minimal intervention, is not directly related to fault tolerance.

When a key management system is outside of the cloud environment hosting the application, availability is a primary concern because any access issues with the encryption keys will render the entire application unusable.