Site Monitoring Modes
A site with more than one monitoring probe can either assign nodes to probes explicitly, or spread the monitoring load across all of its probes automatically.
Atlas SettingsSitesMonitoring
Where a network site contains more than one monitoring probe, you choose how the monitoring work is divided between them. The two modes answer different needs, and you cannot combine them on the same site.
Offline Collection
Nodes are assigned to a specific probe, and that probe is responsible for them.
Because the assignment is fixed, a probe knows what it is meant to be monitoring even when it cannot reach the server — which is what allows it to keep working through a disconnection, buffering results locally and synchronizing them once the connection returns. See Probe Offline Monitoring Mode for what is and is not collected during an outage.
A Data Retention setting controls how long the probe may keep buffering while it stays disconnected. The default is 24 hours.
You choose the probe in the places where monitoring is set up:
- When adding a node
- When adding an IP network in the network monitoring properties
- When adding an Active Directory domain, for the auto-scanner
- Afterwards, in the node settings
Load Balancing
Monitoring is distributed across every probe in the site automatically.
If a probe becomes unavailable, the nodes it was monitoring are taken over by the remaining probes, so monitoring continues without manual intervention. Newly added nodes are distributed automatically to keep the probes evenly loaded.
The trade-off is control: you cannot choose which probe monitors a given node, because the assignment is not yours to make and can change at any time.
Offline monitoring is not available on a site running in Load Balancing mode. A probe that has no fixed set of nodes cannot carry on monitoring them by itself, and workload that moves between probes cannot be buffered coherently on any one of them.
Choosing a Mode
- Offline Collection
- Sites on unreliable links, and sites where you need to know which probe is responsible for a given node.
- Load Balancing
- Large sites on a dependable link, where probe redundancy and even loading matter more than explicit assignment.