Hi guys,
vCOPS (5.7.1) is reporting that I have 100% Network I/O utilization on a few of my hosts. I don't want to say this is impossible, but I have 2x 10G network cables connected and they are LACPed together. Furthermore, vCOPS is reporting that my max Network I/O is 26,136KBps. I'm really confused on how it's getting this number, but that number appears to be different across all of my hosts for max Network I/O. The lowest is ~12K and the highest ~30K. Is this because that threshold is generated dynamically?
I really do appreciate any help, feedback, or direction.
Workload is based on the Demand for resources.
Network IO that you see there is not the actual usage based on the pnic speed and usage but the Demand for Network Resources on the Host/VM/Cluster.
The Demand for Network resources is simplistically calculates the way I explained above.
So you are not seeing usage of the Nic but the demand for network resource on the Nic as a percentage.
One other quick thing to note: With the exception of the VDS, that is backed by a Cisco 1110, we do have one Stand Switch, which is what my TrendMicro and vSheild Edge solution uses.
Hi Chris,
The Collector collects two values of each metric every 5min...and avg and a max value for the metric.
It is this max value that is used to set the 100% of the Network Util.
Example:
Lets say a max value of 200 is collected. Thus if the next avg is 50 then the util is 25% (50 is 25% of 200)
Let say the next avg value collected is 300. Then the Util will be 150% and the new max value is set to 300.
Next collection the avg is 50 again...so now the util is 16.6%
The same for disk IO is valid.
Hugo Strydom
If this is the case how will I know when a Max Capacity alert is triggered if it's accurate? I've been getting these alerts now every few days for different host servers.
Furthermore, is this type of alert most people "tune" out?
You will need to check the usage on the cards...they are located in the : Operations -> All Metric -> Network
Ideally you want a setting to override the max Network setting per host /per cluster/per vm...you can see this becomes complicated.
Ok - Attached is a copy of the transmitted and received graph. What exactly is it am I looking for?
This again just shows me where the traffic appears to have peaked without what the actual MAX is. At the very least it should be 10,000,000 per adapter.
Workload is based on the Demand for resources.
Network IO that you see there is not the actual usage based on the pnic speed and usage but the Demand for Network Resources on the Host/VM/Cluster.
The Demand for Network resources is simplistically calculates the way I explained above.
So you are not seeing usage of the Nic but the demand for network resource on the Nic as a percentage.
Got ya! So I may see a few of these alerts until "normal" changes. I'm clear now. Thank you!
Pleasure!
Correct...if it is a new installation of vCOPs you need to wait at least 2 weeks for VCOPs to "settle" in and learn what is normal in your environment.
I've been using this product now for ~12 weeks and it seems like I add about 40 new VM's every two weeks. I'm sure vCOPS is really confused right now about what normal is suppose to be. :smileyshocked:
vCOPs is not confused...it is showing you correctly that the demand for Network resources is increasing over time ![]()
