The Choice between loading an agent on a managed server and collecting data from an external
vantage point often comes down to a balance between the criticality of the monitored server
and preformance impact of polling.
Agent-based in-band monitoring runs in OS, in parallel with critical applications. In the
event of a hardware fault, the OS and by extension, the agent - may go down, preventing a critical
alert from being sent. However, software issues like application performance problems may be
detected immediately, with detailed alerts then forward to an administrator.
Because out-of-band agentyless monitoring depends on polling, administrator using this option
must decide how often to poll the monitored server. Even if the hardware or OS were to crash,
agentless monitoring would alert the administrator - but only the next time the server is polled.
No comments:
Post a Comment