Active and passive monitoring - agentful vs. agentless

There are two types of monitoring: active and passive.

Active monitoring implies periodical probing devices to define their availability and state of services they provide.

Passive monitoring implies waiting for messages from devices about occurred events. Usually devices send the messages using SMTP Traps.

Both types can be divided to subtypes by the possibility of agent installation to the monitored node or system: agentful and agentless.

Agent is a software app installed on a virtual or physical server or other infrastructure node to implement checks, collect and send monitoring data to the system server. To implement monitoring of any infrastructure at least one agent must be installed (for example, on a server of the system).

In order to learn more about agent installation and configuration refer to the section Agent installation and configuration.

Examples of active/passive and agentful/agentless monitoring combinations:

image
  1. Agentful Active is active monitoring with an agent installed on the monitored host. Agent checks the local HDD and sends the data to the system server.

  2. Agentless Active is active monitoring with an agent installed on a remote host. Agent on the system server checks availability of the website and records the results to the system server.

  3. Agentful Passive is passive monitoring with an agent installed on the monitored host. Agent receives local SNMP Traps and sends the data to the system server.

  4. Agentless Passive is passive monitoring with an agent installed on a remote host. Agent receives SNMP Traps from external system and records the results to the system server.

In some cases receiving SNMP Trap locally and sending data to the system server (example 3) is preferable than sending SNMP Trap to the remote agent (example 4) because SNMP Traps use the UDP protocol, which does not guarantee delivery of messages, while agents establish TCP connection to the server that is more reliable.

In case of agentless monitoring, using of agent installed on the system server is not obligatory. It is possible to use any agent that is available in the current part of infrastructure:

image