From: James Ventre (messageboard@ventrefamily.com)
Date: Mon May 29 2006 - 19:14:05 ART
Other than for the lab, I don't know why you would use load anyway.
It's is a value that is calculated on a 5 minute exponentially weighted
average. It's updated every 5 seconds. That's hardly a sufficient
reaction time to effectively do traffic engineering.
Besides, unless you're in an isolated island, changing the K Value means
you have to change it in a TON of places since the K Values must match
to form a neighbor relationship ... it starts a domino effect.
James
Kemal YILDIRIM wrote:
> As some commands depreciated by Cisco in time, EIGRP metric calculation
> formula need to be handled differently than the original definition.
>
> Actually you will probably not find a good document on Cisco web site about
> EIGRP weights setting. The designers of EIGRP protocol, need to think every
> possiblities for their work, this is why we have never used K values.
> Who wants an unstable network because of unpredictible load that is
> effecting the route metric, also who wants a network that has a mtu
> dependent metric!
> Cisco silently saying that "don't use load, mtu, neither reliability"
> Regards,
> Kemal
>
> Objects are what is unalterable and subsistent;
> their configuration is what is changing and unstable.
> So, you need PRACTICE, PRACTICE, AND MORE PRACTICE...
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> James Ventre
> Sent: Tuesday, May 30, 2006 12:00 AM
> To: CCIEin2006
> Cc: Cisco certification
> Subject: Re: EIGRP K-values
>
> The (Minimum) MTU is not considered in metric calculation. Minimum MTU
> is carried in advertisements though.
>
> James
>
> CCIEin2006 wrote:
>
> Fair enough..but I cannot tell where MTU plugs into the equation...
>
> On 5/29/06, James Ventre <messageboard@ventrefamily.com> wrote:
>
> >Thanks but this does not really tell you that K2 is the metric
> for load
>
> Yes, it does. But it requires you to look at the equation and
> understand it, rather than just memorizing tidbits. Look at the
> snippit surrounding K2.
>
> (K2 * Bandwidth)/(256-load)
>
> If K2 is 0 ..... then outcome of that equation will always be
> ZERO (in fractions .... zero over anything is always zero), which
> means that the load value is never considered. A Non-Zero value
> means the outcome will not be NON-ZERO, thus making a difference.
>
> James
>
> CCIEin2006 wrote:
>
> Thanks but this does not really tell you that K2 is the
> metric for load. I need something like this: K1 - Bandwidth
> K2 - LoadK3 - DelayK4 - ReliabilityK5 - MTU I'm not sure if
> the above values are correct but I would like to have
> something along those lines.
> On 5/29/06, James Ventre <messageboard@ventrefamily.com >
> wrote:
>
> Set K2 to 1 if you want it to consider load.
>
> http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a008009405c
> .shtml#rise
>
> Metric = [K1 * Bandwidth + (K2 * Bandwidth)/(256-load) +
> K3*Delay] *
> [K5/(reliability + K4)]
>
> The default constant values are K1 = K3 = 1 and K2 = K4 =
> K5 = 0.
>
> James
>
> CCIEin2006 wrote:
> > Hello,
> >
> > Is there a quick command to find out what eigrp
> K-values pertain to the
> > different metrics i.e. bandwidth delay reliability load
> MTU?
> >
> > Or is this on the documentation CD? I can't find it.
> >
> > I would like to have a quick reference so that if asked
> on the exam to
> > include load and mtu in the EIGRP calculation I would
> know which K-values to
> > tweak.
> >
> > Thanks!
> >
> >
>
> _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Thu Jun 01 2006 - 06:33:22 ART