Accurate bandwidth utilization

[atef yamin]

Hi all,

I am currently using PRTG for my network monitoring and I rely on SNMP or Netflow.

Having said that, i dont believe that either of above are actually real time (please correct me if I am wrong).

What do you use or what do you recommend for accurate real time bandwidth monitoring. I dont want averages, I am looking for real time data. a spike that takes place only for 500 millisecond (very typical of TCP windowing) thus causing network choke for my voice traffic is not captured by SNMP or Netflow. I believe that there must be a software/hardware that can proxy the traffic and provide accurate stats on it. what do you recommend?

Thanks,

[Vinod Arya]

Atef,

What you want is practically very difficult to get. 

Bandwidth utilization is not available directly. It is based on a formula to be calculated with a couple of essential values/inputs. Check the document on how tools calculate it using SNMP.

First all the CLI data gets synchronised to SNMP DB and than most of the tool poll devices in usualy 5 sec, 10 sec or more intervals.

Unlike interface or other counters utilisation is dynamic and changes frequently. So if some spike stays only for few ms, and device is not polled just in that instance, it may be difficult for tool to track.

However, graphs may show some part of it later.

-Thanks

Vinod

[atef yamin]

Hi Vinod,

you just explainedthe problem. I am looking for the answer :)

How about a software that can proxy the data through. I think riverbed's packet analyzer has the ability to be fed a continuous stream of packets. 

I am looking to get info from someone who has fot this solved.

Thanks

[Alex Pfeil]

We still have this issue today. We see that average utilization, but we cannot see the brief spikes that cause packets to be dropped somewhere along the path. It makes it hard to troubleshoot.