Windows Server 2019 Standard (Virtual Machine)

It is possible, for example when you use NAS.

well, the Server is on Cloud based solution, based on VMware vCloud Director. I assume our storage is a LUN from a shared datastore.

If this is the reason for such behavior, the only solution for me then is to disable these items.

PS: built in Windows performance monitor never registered more than 100% utilization.

I have seen such picture on my cloud storage, I should say that was normal and usual thing for high loaded instances. As I got myself, it is due to the fact how information about utilization is processed. Usually the information is gathered in certain intervals, say 1 second. When you've exhausting disk tasks, the intervals can shift to some milliseconds to get things done thus giving some positive gitter. When Zabbix agent processes these values, it gets more or less data in the interval it considers static so it can lower or higher the actual value. I bet you never mentioned lower ones, but higher are obvious.

I suppose you should pay bigger attention to the average values and discard spikes.

Did you have any extreme values when it is not so loaded?

not so sure. It happens daily, multiple times mostly on a secondary DB server, from where most of the reports and backups are generated.

Still, I don't see this behavior on the primary Database server (same OS, same hw, but different psychical location) and it's heavily used, since it does the main job, but it may not have same impact on the disk as much as the secondary.

Will continue work through https://support.zabbix.com/browse/ZBX-19514