# Traditional Monitoring — Zabbix and Nagios Zabbix and Nagios are the two dominant open-source infrastructure monitoring platforms that most data centers have historically relied on. Understanding what they do well — and where they fall short — clarifies why purpose-built DCIM platforms like Delta's exist. ## What they are **Nagios** (1999) is the older of the two. It works through plugins — small scripts that check whether a service is running, a disk is full, a CPU is overloaded. You define what to monitor and what thresholds trigger alerts. All configuration is done via text files. It's extremely customisable but requires manual setup for everything. **Zabbix** (2001) does broadly the same job but with a web-based UI, built-in graphing, and auto-discovery of devices on a network. Configuration happens through the browser rather than config files. It scales better out of the box and is generally considered easier to operate at scale. Both are free and open-source. Nagios also sells an enterprise version (Nagios XI) with a GUI and additional features. Zabbix is fully open-source with no paid tier. ## What they monitor well - **Server health:** CPU utilisation, memory usage, disk space, process counts - **Network devices:** Switch port status, bandwidth utilisation, latency, packet loss - **Services:** Is the web server responding? Is the database accepting connections? Is DNS resolving? - **Custom checks:** Both support writing your own monitoring plugins for anything that can be scripted — application-specific metrics, API health, log file patterns They're essentially "is this device/service alive and within normal parameters?" tools. They poll at intervals (typically 30-300 seconds), collect metrics, store history, and fire alerts when thresholds are breached. ## Where they fall short for data centers **No physical awareness.** Zabbix knows server-07 exists and its CPU is at 80%. It has no concept of *where* server-07 physically sits — which rack, which U position, which row, which room. That spatial awareness lives in a separate spreadsheet or CMDB that someone maintains manually. **No power or thermal context.** Traditional monitoring tools don't natively understand rack-level power draw, per-U thermal contribution, or cooling capacity. You can bolt on IPMI checks and SNMP traps for environmental sensors, but the data lives in isolation — it's not correlated with physical location or capacity planning. **No capacity planning.** If you want to know "can I fit a 4U GPU server drawing 3kW into rack 12?", Zabbix can't answer that. You need to cross-reference the asset spreadsheet, the power monitoring system, and the thermal model separately. **Scale of configuration.** In a large data center with thousands of devices, maintaining Nagios config files or even Zabbix templates becomes its own operational burden. Every new server type, every firmware version change, every new sensor requires template updates. ## What DCIM replaces A purpose-built DCIM (Data Center Infrastructure Management) system like Delta's collapses several separate tools into one: - Infrastructure monitoring (Zabbix/Nagios replacement) - Physical asset tracking (CMDB/spreadsheet replacement) - Power monitoring (PDU/metering replacement) - Thermal mapping (CFD/sensor replacement) - Capacity planning (manual calculation replacement) The key difference is that DCIM understands the *physical topology* of a data center — not just the logical network topology. It knows that server X is in rack Y at position U23, drawing Z watts, and contributing W degrees to the thermal profile of that rack section. ## When traditional monitoring still makes sense Zabbix and Nagios aren't obsolete — they're just the wrong tool for physical infrastructure management. They remain excellent for application-level monitoring, network service health checks, and environments where the physical layer is someone else's problem (cloud, colo where the provider handles facilities). Many organisations run both: DCIM for the physical plant, Zabbix/Nagios/Prometheus for application and service monitoring. --- See also: [[Delta DCIM — U-Level Granularity]] | [[Modular Data Centers MoC]]