Change Control Practices

Once the production system platform is stable and all applications are fully operational, any proposed system changes must go through a peer review to identify the impact of the changes and to provide a strategy for implementation.

Rebooting a system immediately after implementing any changes is recommended. This action helps to avoid latent configuration change errors such as typographical errors in the /etc/system file. Associating a failed reboot at some future time with previous modifications can be extremely difficult.

Data centers with higher availability requirements should have a production system mirrored by a test system (the mirror could be a scaled-down version of the production system) to implement and evaluate any changes before adoption. In some environments, the development system can also be used as a test system. Take care to isolate systems administration operations on the test system from the real production environment. An example of a system administration operation error: A customer mirroring an SAP manufacturing environment inadvertently generated a test production order, which resulted in the building of a $1.5 million piece of equipment.

Backing up a production system to tape or other persistent storage before you implement any changes is imperative. This backup enables restoration of the original system if the proposed modifications prove detrimental.