Suppose a network consisting of hubs A, B, C, D ... and X. I am looking for ways to visualize how users use the network such as incoming, outgoing and other metrics. In Apache logs, I can see some errs if something did not work but I have no realistic picture about such a system in general i.e. how the system actually works. I am looking for some sort of flow-analysis and I would like to get pure data to create some graph. Then analyze the graph with some metrics where I do not even know the right metrics, perhaps some dispersion metric. My goal is to create some sort of objective way to judge quality.
etherape? mrtg? argus? pastMon? How much data are you gathering, how long do you want to store it? You've not given a lot of information.
However network monitoring is not the solution to diagnosing application faults. Maybe you're looking for mod_security.
I don't think your question is specific enough to actually answer. Here's a couple things to try and why:
- Boundary. Collector sends granular stats to Boundary, where you can see the impact of traffic flows and ideally, identify anomalies.
- Scout. Automatically measures per-server network stats and makes it really easy to write a custom plugin of your own.
- Blitz.io. If you're mostly focused on HTTP, this will at least let you discover how your site degrades under load.
There isn't a single "objective way to judge quality," but based on your question, any or all of these will be a huge improvement from where you are.