Logging should be configured such that: Logging level is set to a level sufficient for the target device Logs should be sent off the device to a syslog or trap server or servers Logs should be sourced from a consistent interface to ensure easy attribution of logs to the correct device Logging levels should be explicitly set to a level appropriate to the device. Rationale: Logging on any network device is always limited by how much storage can be set aside for logs. It's important for this reason to send all log entries to a central device that can collect and correlate all logs, either in a database or in flat text files. The key thing this approach contributes is central logs on a larger storage device (disk) Logging to an off-device target also makes clearing any incriminating logs more difficult for an attacker, or if an attempt is made to hide a mistake. Logging off-device also ensures that any clearing of logs is also seen and can be alerted on. Sourcing all logs from a consistent interface ensures that log entries can be easily attributed to the correct device once they arrive at the log server. If a logging interface is not set, the source IP address of individual log entries can change as the network topology changes. This situation can make any subsequent log analysis more difficult. Impact: Because syslog traffic is not encrypted, it's recommended to ensure that the path the log traffic takes is not susceptible to any MiTM (Monkey in the Middle) attacks. Often this means assigning a dedicated management interface, which by default is in a separate VRF.