[mrtg] Minimum hardware

McDonald, Daniel (Dan) Dan.McDonald at austinenergy.com
Thu Jun 2 17:08:09 CEST 2016

From: mrtg <mrtg-bounces+dan.mcdonald=austinenergy.com at lists.oetiker.ch<mailto:mrtg-bounces+dan.mcdonald=austinenergy.com at lists.oetiker.ch>> on behalf of thomaz portella <thomazp at gmail.com<mailto:thomazp at gmail.com>>

I am reading the necessity of using RRDtool together with mrtg.

Yes.  Rrdtool scales much greater than rateup.

Which benefits will I have using it or not ?

Greater granularity (up to millisecond precision if you were to poll that often) as opposed to fixed at 5 minute polling cycle.
Support for negative numbers (for example, temperature sensors or dB meters)
Faster update times - just updating a database that is designed to consolidate daily/weekly/monthly/yearly views rather than fiddling with flat files that take a long time to parse and rewrite.
Less up-front processing.  Rateup creates a graph on every polling cycle.  Rrdtool just updates the data and lets you create the graphs at time of view, leading to much lower cpu utilization.
More flexible graphs - rrdtool gives you almost limitless formatting capabilities and combined metrics, while rateup just creates one style of graph.

I need an external database or could be embedded ?

Not for rateup.  You may want one if you are tracking 100’s of switches.  Large sites generally set up mrtg configs for each switch as a separate file, and then include them all from a master mrtg.cfg file that just contains include statements from all of the switches.

For example, I have about 700 devices, all with unique snmp credentials.  I have a database that lists the credentials, as well as which template should be applied to each device.  I have a script that will check to see which devices are reachable, create a mrtg.cfg file with all of those config files included, and also put together a script to run mkconfig for a subset of the devices that are being managed.

Tks ,

On Jun 2, 2016 11:16 AM, "thomaz portella" <thomazp at gmail.com<mailto:thomazp at gmail.com>> wrote:

We don't need to monitor these servers the number I sent.
We need to follow the network elements(switches and routers) behaviour . We have approximately 100 of this . I couldn't sent this number before because I didn't have it.
I have to know an idea of the server I need, to ask here

On Jun 1, 2016 6:57 PM, "thomaz portella" <thomazp at gmail.com<mailto:thomazp at gmail.com>> wrote:

2500 RH boxes:
1000 physical
1500 Virtual
When I asked, the person could tell me how many switches was out, but I can ask this tomorrow.
Any other question will be answered if necessary.

On Jun 1, 2016 5:59 PM, "Wm. Josiah Erikson" <wjeNS at hampshire.edu<mailto:wjeNS at hampshire.edu>> wrote:
2500 Red Hat boxes or 2500 switches? Do you mean analyze or just poll traffic data? Please describe your environment and exactly what you want to analyze more thoroughly.

On June 1, 2016 4:23:13 PM EDT, thomaz portella <thomazp at gmail.com<mailto:thomazp at gmail.com>> wrote:

I need to configure a mrtg server to analyze my network.
We have approx 2500 machines, and use Red Hat.
Any other question, I can ask here.
I need to know the minimum hardware needed.


mrtg mailing list
mrtg at lists.oetiker.ch<mailto:mrtg at lists.oetiker.ch>

Sent from my Android device with K-9 Mail. Please excuse my brevity.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.oetiker.ch/pipermail/mrtg/attachments/20160602/a392e35c/attachment-0001.html>

More information about the mrtg mailing list