I've been using Observium for a while and there seems to be some discrepancies between my graphs and the upstream (my data center provides me Ethernet GigE bandwidth). For example, from around 8am to 12pm today it showed overall utilization pretty consistent well above 400Mbps. I have two drops from them: primary and secondary (redundant). Plus, they both show active instead of active/passive. Not sure if that has anything to do with it.
From the tech:
"95th percentile is calculated from both ingress and egress traffic. For example, if you used a constant 200Mbps up and down, then you would only be charged your commit of 200Mbps, not double. I do see what you mean that the two handoffs (layer 2) are equaling more than 300Mbps, but when you look at the Aggregate (layer 3) graph at the top, it does not go over 300Mbps. These packets can possibly be queued before going out so you don’t go over your commit. "
So, what do you think might be causing this hiccup?
I noticed this too: so these switches are actually trunked (2 x 3650 L3 switches). In the graph, it also shows the uplink connectivity calculated with the total utilization. This is definitely not accurate as it shouldn't do that.
Example:
Drop #1 = 40Mbps
Drop #2 = 200Mbps
Uplink to Switch #2 = 130Mbps
Now as a result VLAN2155 (total utilization) is calculating not only #1 and #2 drops from the upstream, but even the trunk port labeled "Uplink to Switch #2".
How can I rectify this with Observium to avoid that and get an accurate output?