Graphs showing wrong usage - Update, Please look -

[rjoshuaw]

Hello all!
So I've started back on the Cacti track to get it up and running at my business and added a few devices to test out. One of my tests was to compare what it shows as bandwidth utilization compared to Orions Network Performance Monitor. I've yet to look through all of the devices this morning, but I've attached an example below

This is a Cisco 3560-48-PS I'm using for a core at another building. It has a Gig link back to a 6509 via fiber. It also has gig links to the other floors in the building. I'm using the interface graph that comes with Cacti, we're on version 8.7e. I wanted to make sure it was working before trying to upgrade to the new version. Any ideas? I'm going to start look at the other devices and will add those as well as I find them.

BTW: I am using the 64byte graph

Thanks,
Raun


I've attached an extreme example as well. This is a 6509 with a Ten Gig connection to a cabinet 4948 switch.

[jay]

Hi

It looks like you Orion graphs are bits per second where as you are using the bytes per second graphs in Cacti. so 2MBytes per second equates to 16mbits per second. This is why your graphs look so different. The bottom 2 graphs show this the best as the output is in a nice easy to read line. Try using the bits per second graphs in Cacti. You should then see similar results between Cacti and Orion .

Hope this helps.

Cheers

Jay

[rjoshuaw]

Thank you for the reply. I feel like an idiot, I just figured this out a bit ago and haven't had a chance to update the topic. Grrrrrr.. Again, thank you. It's nice to have a community of support. ----------------------

Well, perhaps it's not completely straight after all. Take a look. This is a 64bit graph. The desperaty increases in usage it seems. 6509 with 10gig link to a 4948

[rjoshuaw]

Any ideas.

[jay]

Hi

What is the maximum value set to in the data template for this graph? Maybe the figure is set too low.

Cheers

Jay

[rjoshuaw]

Thanks for the reply,
The current value for both input and output is set to: 10000000000

-Raun

[jay]

Hi

Remember that Cacti will only plot the average on the graph, unless you tell it otherwise in the graph template. I would still expect to see the maximum as being nearer what you see in Orion.

Have a read through the pages on rrd http://docs.cacti.net/manual:087:8_rrdtool#rrdtool it also shows you how to plot a maximum figure in your graphs. Might be useful for you as you are comparing Cacti stats with Orion, which plots the maximum figure.

What are your polling times set to for Orion and Cacti. It looks like Cacti is set to the standard of 5 minutes. Is Orion set to 10 minutes?

As you are looking at the first days data I wouldn't expect consolidation to be causing the figures to be out.

Maybe someone else can jump in and tell him why the graphs/figures aren't nearer to each other.

Cheers

Jay

[rjoshuaw]

Jay, I think you're sending me down the right path. With the MAX added to the graph it was more consistant to the Orion version. Are you aware of any graphs for Cisco Interface's that some one else has built along these lines? I ask because I know there are lot more talented people out there then me simply fumbling around and they'll probably look nicer too. Was searching the forum, but now the search says it's not available. heh

[jay]

Hi

Cool. I'm in the process of having to reconfigure my MAX Ds to 1G as I have seen issues with spikes in my graphs.

The standard interface graph is all you need to monitor your Cisco interfaces. It just uses the standard MIB. I use the standard template to monitor our Cisco routers and Extreme switches.

You can add additional information to the graphs using comments to show total data and other information. You can also set the graphs to show input as positive and output as a negative. The rrdtool links I sent will show you what I mean.

Have a look here for inspiration

http://www.mrtg.org/rrdtool/gallery/index.en.html

Regarding your issues what settings do you have in your graph template regarding autop scaling. Might be this thats causing an issue.

Cheers

Jay

[rjoshuaw]

I think we have this settled down. I've actually done away with my old cacti and rebuilt a fresh install with 8.7g. I'm graphing ok right now, but my max is not showing up for some reason:



Also, what do you mean by spikes. Is this simliar to what I'm seeing here? How does one distinguish from reguler traffic versus what your describing as a graph here? What are maninpulating to resolve it?

[mahade047]

I have created aggregate graph with 3 different graphs.
the records under the graphs are showing correct data [7.5+17.6.19.1 = 44.2G]of individual graphs
but the aggerate graph in showing 40G max where it should be 44+ G [photo attached]

what mistake I made here ?