Zabbix

What Zabbix does

Zabbix is one of the oldest open source monitoring systems still in active development, and it is built to watch whole IT estates rather than a handful of web services. A central server gathers data, evaluates it against triggers and raises problems; a PHP web frontend handles configuration, dashboards and reports; and a relational database stores both configuration and history.

Data comes in through many routes, which is Zabbix’s main strength. The Zabbix agent, or the newer agent 2 written in Go with plugins for services such as Docker and databases, runs on Linux, Windows, macOS and several Unix systems. For devices that cannot run an agent there are SNMP polling and traps, IPMI for server hardware, JMX through a Java gateway, SSH and Telnet commands, ODBC database queries, HTTP requests and scraping of Prometheus endpoints. Web scenarios run multi-step HTTP checks, and browser items drive a real browser through a WebDriver for synthetic website tests. Discovery rules find new network devices, hosts and resources such as disks or interfaces and apply templates automatically, so large fleets do not have to be configured by hand.

Alerts and reporting

Triggers can depend on one another, so a failed switch does not produce an alert for every server behind it, and event correlation helps narrow down root causes. Actions send notifications through media types, with escalation steps when a problem is not acknowledged. The repository includes ready-made integrations for Slack, Discord, Mattermost, Telegram, Pushover, PagerDuty, Opsgenie, Jira, ServiceNow, Zendesk and other help desks, plus email and SMS. A services tree calculates SLAs, and scheduled reports, network maps and geomaps help when presenting availability to others.

Self-hosting Zabbix

Zabbix is installed from official packages for the major Linux distributions, from source, from official container images or as a virtual appliance. The server and proxy run on Linux and Unix only. The database can be MySQL or Percona 8.0.30 or later, MariaDB 10.5 or later, or PostgreSQL 13 or later, optionally with TimescaleDB, and the frontend needs PHP 8 with Apache or Nginx. The documentation’s sizing examples start at 2 vCPUs and 8 GiB of RAM for about 1,000 monitored metrics and 4 vCPUs and 16 GiB for 10,000, and it recommends a separate database server for large installations.

For distributed environments, Zabbix proxies collect data in remote sites or behind firewalls and forward it to the server, and a native high-availability mode runs several servers with one active and the rest on standby. Releases follow a published lifecycle with long-term support versions.

License and hosted version

Zabbix moved from GPL-2.0 to AGPL-3.0 with version 7.0. There is no paid edition of the software: Zabbix SIA earns money from support subscriptions, consulting and training, and asks commercial users to consider buying support. Zabbix Cloud is its managed service, with a choice of five data-center regions, tiers sized by the rate of incoming values, a short free trial and cloud proxies for remote networks. Note that the documentation itself is not openly licensed.

Limitations

Zabbix has many concepts to learn: hosts, items, triggers, templates, actions and media types all need to be understood before the system behaves the way you want. History grows quickly in the database, so housekeeping and sizing need attention. It does not do distributed tracing or application performance monitoring, and its log monitoring is item-based rather than a log search platform.

Who it suits

Zabbix suits system and network administrators who must monitor mixed infrastructure, including switches, routers, storage, hypervisors and Windows servers, from one place, and organizations that want a long-supported product with commercial backing available. Cloud-native teams focused on Kubernetes and microservices often prefer Prometheus with Grafana, or a combined tool such as SigNoz.