In the last column, I explained open-source consolidated monitoring system Zabbix’s external and internal monitoring functions and its other monitoring functions. In this installment, we’ll look at hosts and host groups, which contain monitoring items.
Monitoring items can be managed for each host. What’s more, hosts can be managed together as a host group.By using hosts and host groups, you can easily understand the targets being monitored.
A host manages one server (either virtual or physical). It contains the items to be monitored (monitoring items). So one host has multiple monitoring items. The monitoring items don’t distinguish between physical and virtual machines. They become information gathered in general by Zabbix Agent. In short, this is white-box monitoring.
At the host level, CPU/memory/storage/network loads can be monitored.
Hardware appliances that don’t allow Zabbix Agent to be installed, such as network and storage devices, are managed with SNMP. For items monitored with SNMP, one device is considered one host.
Hosts can be organized into a host group. By handling hosts collectively like cattle, you can conveniently group them together to monitor one service. Bringing anomalies of a service together into a group is useful for maintaining service continuity.
For microservices that only execute one service per host, creating a host group for each service is also helpful. Create a host group for Web servers, application servers, DB servers, storage, and network appliances. By increasing/decreasing the amount of services needed for each host group and its corresponding resources, you can visualize service continuity.
Zabbix can use templates, which are sets of multiple monitoring items. The following are examples of templates.
- Operating system (OS) templates
- Mac OS X
- Windows Server
- Network devices
- Service monitoring
Multiple templates can be assigned to a host. In short, you can monitor a host by assigning an OS template and templates for different services.
When using an OS template, items differ for each host. Take, for example, the items below.
- Network interfaces
- CPUs and CPU cores
- SNMP OID
- JMX (Java Management Extensions) object
- ODBC SQL queries
- Windows Services
- host interfaces
Zabbix has a function called low-level discovery. It allows the items above to be automatically detected. The automatic search creates items, which are registered as items belonging to the host.
Low-leveldiscovery is included in OS templates and SNMP templates. When OS and SNMP (for each device) templates are assigned to a host, host-specificinformation is automatically registered.
Automatic host registration
Install and configure Zabbix Agent when registering a new host. Afterwards, when Zabbix Agent connects to Zabbix Server, the host can be automatically registered.Conversely, when communication with Zabbix Agent is lost, the host can be automatically removed.
For automatic host registration, use Zabbix Agent settings or the uname command to automatically identify the host name and OS. An OS template can be automatically assigned after the OS is determined. Hosts can also be automatically assigned to host groups.By using automatic registration, you can flexibly deal with the changing number of hosts in a cloud environment, which contain a vast number of hosts.
Network discovery is a Zabbix function that detects and registers services on a particular network.Its prerequisite is automatic host registration (set by Zabbix Agent configuration).Network discovery automatically detects the following network services and adds corresponding service monitoring.
- General service protocols
- TCP port connection
- Information gathering by Zabbix Agent
- Information gathering by SNMP agent
- Ping response
You can also specify the host’s network range (e. g. 192.168.200.0/24) for discovery.
With this function, templates (SMTP/Web server, etc. ) needed for monitoring not only the operating system but also started services can be automatically added. Each function can also be made a host group.
VMware and VMware ESX comprise an outstanding virtual computing platform. Zabbix supports VMware.It can automatically discover hypervisors and virtual machines.
These virtual machines are registered in the form of hosts for monitoring by using templates.
Virtual machine information is gathered from the VMware collector launched from Zabbix processes. VMware configuration data and VMware performance data are collected separately. It is thus recommended that you install more VMware collectors than the number of VMware services being monitored.
Currently, only datastore, network interface, and disk device statistics and custom performance counter items can be collected from VMware performance counter information.
In this column, I explained hosts and host groups, which are the units of monitoring by Zabbix. You can reduce the work involved in adding new hosts by skillfully using hosts, host groups, and templates. You can also use Zabbix’s network, low-level,and VMware discoveries to easily deal with the changing number of hosts. Having a good command of these functions is critical for making full use of Zabbix.
In the next installment, I’ll talk about actions to take after monitoring. See you then!