Help with UNDER-Utilization monitoring

[delija91p]

Hey you guys, I do have an earlier post about a trigger that would be able to help me monitor over a period of 7 days of a server's CPU Utilization being under 10 percent. https://www.zabbix.com/forum/showthread.php?t=48339 The issue is that giving me an average over 7 days, is very memory intensive, and my Zabbix server is not just a Zabbix Server, I have it doing other things. So I was wondering if anyone had any ideas and or suggestions on how I should approach this issue... Is there a smarter way to monitor Under Utilization. The reason I want to monitor it over 7 days is because I want a weekly email of all the servers that have been under utilized that week, that way instead of getting in new servers, I can utilize those servers.

Thank you guys in advance,
I have been struggling with this for quite a bit.

Also this is my suggestion for the next version of Zabbix. I feel like many people are monitoring over utilization of servers. But it would be nice if it was easier to monitor under-utilization for people who have more servers, and we need to verify which servers are being used and under utilized. So maybe instead of storing the cache values for each server in memory, maybe storing it on the hard disks would be an improvement that you guys could make??

Stephan Stankovic

[BDiE8VNy]

How about using a precalculated average by a calculated item as a work-around?
This way one has control via its update interval about how many values get stored in value cache.

Item
Name: Average CPU idle time ($4)
Type: Calculated
Key: calc.cpu.util[,idle,avg1,1h]
Formular: avg("Template CPU Usage:system.cpu.util[,idle,avg1]",1h)
Update interval: 3600

Trigger
Trigger expression: {Template CPU Usage:calc.cpu.util[,idle,avg1,1h].avg(7d)} > 90

Value cache (1 system.cpu.util value/min)
(60 of agent item values) + (24 * 7 of calculated item values)
or:
228 vs. 10080 per host cached values

Not tested though.
Something similar to ZBXNEXT-2463 might be imaginable too.