How to use the monitoring system correctly?

Question

Monitoring

Good day!
There is a Zabbix 2.2.9 monitoring system and services that require monitoring
Nginx, php-fpm, mariadb, php-apc, redis, wowza, and CentOS 7 itself
Questions in order:
1. Where can I find in-depth and clearly stated information about what kind of counter values ​​can I get from these (and possibly not only) services and a description of what these counters are, ideally compiled in one place and up to date. (Russian)
2. How to correctly analyze the data obtained during monitoring:
a) Which values ​​increase means a performance drop, which increases (perhaps some non-trivial cases, not only LA, memory etc)
b) Which counter values ​​are related to others and how
c) "Bad" values ​​of which counters are corrected by service configs, which by changing hardware (or adding / changing technology), which by developers (programmers)
3. Is there a material describing step by step the process of creating your own template for Zabbix (Russian)?
Perhaps I misunderstand what tasks monitoring solves in general, but I want to get a tool that will help me determine if my experiments with configs are beneficial, give programmers graphs that will show whether their optimization experiments are beneficial, and a tool to determine the "bottleneck"

Answer 1

1. Alas, but only on the off site and most likely in English for the time being (considering your version of Zabbix). Unfortunately, it is not always clear from the off-line documentation what is why and where, so good luck to you :)
2. It's too difficult to say like this, not knowing what kind of system you have there in general and how it is connected with what. As an option, turn on everything that is possible if possible (if the power allows) and then filter out the unnecessary.
3. Try to watch these lessons (in Russian, but the sound is very quiet, turn the speakers =)), maybe it will help you somehow: www.youtube.com/playlist?list=PLcDkQ2Au8aVMbBQNjiQ...
P.S. As the comrade said above, if you only need to collect information, then it is better to use something simpler (Munin, or quite on the edge of Nagios), rather than Zabbix.

Answer 2

1) type zabbix’s team in google and go offsite with a list of templates
2) This is a question for you, someone is 20 normally, someone is a lot, most of the triggers are already configured.
3) counters are not connected with others, they are separate entities in essence
4) no counters are
corrected like that 5) the official Zabbix website to help you there is completely Russian documentation
you don’t understand the essence of monitoring
graphics correctly it’s easier to assemble some kind of munin out of the box and not bathe