Nagios

Nagios Core has monitored servers and network devices since 1999, executing scheduled checks written in any language that returns standard Unix exit codes. Its C engine evaluates host hierarchies and dispatches threshold alerts across private networks without sending telemetry to external endpoints.

Why Nagios instead of Datadog

Datadog packages infrastructure monitoring into a managed SaaS platform, billing monthly per host while charging extra for custom metrics, traces, and log volume. Nagios Core handles the foundational availability jobs that keep infrastructure alive: ping checks, port responsiveness, disk capacity, and daemon health. It does not provide distributed tracing or automated cloud telemetry; it simply evaluates check exit codes against warning and critical thresholds.

The cost equation diverges as server fleets expand. For engineering teams running hundreds of bare-metal nodes or private hypervisors, Datadog's per-host model produces substantial annual bills purely to confirm machines are reachable. Nagios Core runs on a single internal virtual machine, polling thousands of endpoints without license fees, seat charges, or egress costs. In exchange, teams manage flat configuration files by hand, deploy local agents like NCPA, and provide their own time-series dashboards.

What it does well

• Plugin interface driven by standard exit codes (0 for OK, 1 for WARNING, 2 for CRITICAL, 3 for UNKNOWN), letting engineers write probes in Bash, Python, or Go.
• Parent-child host topology modeling that distinguishes between a down host and unreachable nodes behind a failed switch, preventing alert storms.
• Remote check execution through NCPA and NRPE for querying internal performance metrics on Linux, BSD, and Windows machines.
• Event handlers that execute custom remediation scripts automatically upon state changes, restarting crashed daemons before paging an engineer.
• Passive check processing via NRDP and NSCA, accepting pushed XML/HTTP results from firewalled subnets.
• Lightweight C daemon that schedules thousands of recurring service checks on minimal CPU and memory.

Limitations

• Configuration reloads require reparsing all flat text files, which can cause defunct processes or fork exhaustion under high check volumes.