I actually did a number of things with RTR one of them is present at the
link below as part of my initial work for my thesis , some of the rest are
still in 'development' in Intel and cannot be seen as it is inside the
firewall until I install them as well in the University but here goes ...
First - I do not claim to be a definitive authority on RTR and I did not
install and test IPM from Cisco as yet so take the following as
observations:
The RTR MIB history entries seem to be based on hourly entries, the number
of hourly entries kept is governed by the configuration
hours-of-statistics-kept , so the easiest thing to do with RTR if you want
to track long range data instead of 5 minutes or so is produce results based
on the hour statistics tables, I am attaching one report I am creating based
on data collected Via SNMP this way and stored in an rrdtool database, it is
in actuality a Grade Of Service report based on grading response time and
packet loss on a few links, I am removing the legend of the links and the Y
scaling so as not to reveal Intel specific data.
Another thing I do is use MRTG to collect the data, the problem is that I
wanted to configure collection often (every minute or less) but collect
every 5 minutes so I choose to get the summary data every 5 minutes and
basically get 5 minutes averages from the hourly counters, I used a script
to find the most recent hourly sum data and return that counter, the problem
is when an hour is rolled, I do not handle that well and in fact one data
point every hour is lost, I intend to work around that later on this month.
Another option is to collect results via SNMP at the same rate as the RTR
check, but if I use a query every 12 seconds (which I would like to do) that
will not be too good.
I am including the MRTG config files as well as the simple scripts used to
get response time and error (actually drop) rates
I hope you find this of value but as I indicated there may be a better ways
to get at this.
One suggestion I can make is this - keep the last value apart for the
collection scripts and process the delta values saving them as GAUGE, when
there is a roll over sum the final value of the closed hour with the current
one to get the delta for this round then continue , I will probably be
writing this in the coming days and can send this as well but for now 11/12
entries are OK.
Now finally - attached please find
1. Scripts to collect data (the .pl files) and configuration file (.cfg)
Example graphs (.png) from the GOS system - legends and Y scaling where
removed so as not to reveal any Intel data, I am calculating the total grade
based on the grade of 7 links, for each the drop rate and the response time
are graded and a total Grade calculated, the colors of the links (or links
data) are consistent in the three PNG images
-----Original Message-----
From: Hank Nussbacher [mailto:hank@att.net.il]
Sent: Fri, November 03, 2000 9:53 AM
To: ; omer.ben-shalom@intel.com
Subject: RE: [nsp] Response Time Reporter comments sort
At 09:32 03/11/00 +0200, you wrote:
Omer, can you answer him...
>Hi there, I am very interested in implementing some RTR trending with MRTG,
>and I really like the look of what was done on your pages.
>
>Could I ask if you would be able to send me the .cfg file(s) you use for
the
>RTR pages please?
>
>Thanking you in advance.
>
>Paul
>
> > -----Original Message-----
> > From: Hank Nussbacher [mailto:hank@att.net.il]
> > Sent: 02 November 2000 21:34
> > To: roliver@chello.com; cisco-nsp@puck.nether.net
> > Subject: RE: [nsp] Response Time Reporter comments sort
> >
> >
> > At 20:02 02/11/00 +0100, Rod Oliver wrote:
> >
> > Yes. See:
> > http://132.67.120.62/mrtg/rtr/
> >
> > -Hank
> > PS I didn't do it. Someone else.
> >
> > >Has anyone had any success with getting MRTG working with RTR?
> > >
> > >Rod Oliver
> > >
> > > > -----Original Message-----
> > > > From: Imre Fitos [mailto:imre@one.net]
> > > > Sent: Wednesday, 1 November 2000 2:16 p.m.
> > > > To: 'cisco-nsp@puck.nether.net'
> > > > Subject: RE: [nsp] Response Time Reporter comments sort
> > > >
> > > >
> > > >
> > > > I've checked it out about a year ago, I liked the idea but the
> > > > software was
> > > > crappy and didn't have enough functionality. As the real work is
> > > > done by the
> > > > routers I was thinking about writing my own software to analyze
> > > > the results.
> > > > Of course I also thought Cisco's netflow analyzer sucked.
> > > >
> > > > Wonder if other SLA packages like Netcool or HP's
> > Vantagepoint support it
> > > > yet?
> > > >
> > > > Imre
> > > >
> > > > From: Rod Oliver
> > > > > I have been looking into RTR with great interest. Does
> > anyone have any
> > > > > comments/advise on its implementation?
> > > >
> > > >
> >
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:20 EDT