I imagine that I am misunderstanding the Zabbix web monitoring model... but why would an application only be able to "contain" one host? I can see why a single host might be used for multiple applications; however, most applications are sufficiently complicated to span multiple hosts.
I'm struggling to really understand what Zabbix's model is for a distributed web application where each single web application (not necessarily a Zabbix "application") might have many hosts.
For example, a LAMP stack containing a replicated MySQL - so effectively two symmetric applications in two data centers with a replicated database between them. I might expect to create two "Host Groups" (one for each data center) and the physical (or VM) hosts respectively within those Host Groups. Then I might expect to create an "Application" with a number of web monitoring scenarios for the hosts. I'd like to do this in a way that doesn't involve installing Zabbix agents everywhere.
Could somebody layout the suggested Zabbix object hierarchy to model/build this "properly?"
TIA
I'm struggling to really understand what Zabbix's model is for a distributed web application where each single web application (not necessarily a Zabbix "application") might have many hosts.
For example, a LAMP stack containing a replicated MySQL - so effectively two symmetric applications in two data centers with a replicated database between them. I might expect to create two "Host Groups" (one for each data center) and the physical (or VM) hosts respectively within those Host Groups. Then I might expect to create an "Application" with a number of web monitoring scenarios for the hosts. I'd like to do this in a way that doesn't involve installing Zabbix agents everywhere.
Could somebody layout the suggested Zabbix object hierarchy to model/build this "properly?"
TIA