From: Eggert, Scott (scott.eggert@berbee.com)
Date: Mon Feb 11 2008 - 19:40:17 ARST
This explanation fits all the pieces together. This really helped. If
you separate out the DLY and BW based on "eigrp" values, it is easier to
understand and remember. Thanks.
________________________________
From: Herbert Maosa [mailto:asawilunda@googlemail.com]
Sent: Monday, February 11, 2008 3:18 PM
To: Eggert, Scott
Cc: ccielab@groupstudy.com
Subject: Re: EIGRP variance load sharing help
I dont think you can avoid the FD as a route will not be considered as
candidate for Load Sharing if it does not meet the feasibility criteria.
I always do the maths on paper ( I really dont know if there are short
cuts ), but I am quite comfortable with this so it does not really eat
up any time for me. If you are finding yourself trying so many different
values, it could be that you need a more solid base on the whole concept
altogether.
Basically, this is how I approach this.
* Multiply the FD by X, where X is the traffic share count being
sought. This gives me what the worst metric should be for the load
sharing to work according to the given ratios.
* Work out X = BW(eigrp) + DLY(eigrp). The goal here is to find
the value for DLY(eigrp) as I never want to touch the BW. I only touch
the BW once in the path, as only the minimum bandwidth for the whole
path is what matters. So I often make the minimum bandwidth for the
worst path to be equal to the minimum bandwidth for the best path. That
way I know that the only discriminating factor is going to be the DLY.
So, re-arranging work out DLY(eigrp) = X - BW(eigrp).
* Once I find the value for DLY(eigrp), I can distribute it
whichever way I want on the links of this path, for as long as at all
times I am achieving the feasibility criteria.
At all times, remembering that DLY(eigrp) = Interface configured Delay *
256
BW(eigrp) = (10^7/interface bw in Kbps)*256
Also, remember that as you do the calculations, round down all real
values ( floating point numbers ) as Cisco routers don't do floating
point maths and they round down at each step of the calculation.
regards,
Herbert.
Herbert.
On Feb 11, 2008 5:55 PM, Eggert, Scott <scott.eggert@berbee.com> wrote:
Can anyone offer any insight on how to quickly determine the EIGRP
metrics for a particular variance load sharing ratio. I struggled with
the IE lab 8 EIGRP variance question and ended up trying different
delays and bandwidth and ultimately calculating them to figure out the
question. However, this could eat up time if this was the real test.
Is there a quick rule or procedure for getting a particular metric and
load sharing ratio? In particular, Is there a way to avoid going over
the FD?
Thanks.
Scott Eggert
Network Engineer
CDW Berbee
4321 W College Ave Ste 400
Appleton, WI 54914
920.996.3014
920.284.7619 - mobile
920.997.9419 - fax
www.berbee.com
This archive was generated by hypermail 2.1.4 : Sat Mar 01 2008 - 16:54:48 ARST