Hey Team,
Bryan R asked a good follow up question and I should add this to this
thread. Many thanks Bryan.
As mentioned above, the variance command instructs EIGRP to install routes
into the routing table with unequal metrics according to the variance
multiplier. The load sharing is not 'smart just yet, but both links will be
used. I mentioned by this being proportional by default, and I was
thinking of the traffic share balance command.
The traffic share balance command 'balances' the traffic across the multiple
paths according to the metric; sent proportional over each link.
This brings up 'traffic share min' which is different, not the default for
traffic share, and less preferred IMO. With this command, traffic will only
be sent via the best or min cost link. Other routes are still put into the
routing table too. Tthis decreases the convergence time in case of a
failure, but does not load balance over the multiple links.
Dominik - CEF is still enabled even if you choose different methods. See
the link I sent earlier for config examples doing that.
HTH Dominik and team,
Andrew Lee Lissitz
On Wed, Apr 22, 2009 at 11:21 AM, ALL From_NJ <all.from.nj_at_gmail.com> wrote:
> Hey Dominik and team,
>
> Interesting question.
>
> EIGRP will calculate the paths and choose the best one. If multiple equal
> cost best paths exist then EIGRP can choose more than one. Nothing new or
> special about this. So when you have multiple paths, unequal ones, you can
> tell EIGRP to use these other paths also. The variance command as you
> mention below will allow these to be installed to the routing table.
>
> So how does CEF learn the next hops and prefixes? It learns it from the
> routing table. So there is a relation to the routing table, routing
> protocol, and CEF.
>
> CEF uses a load sharing table and distributes the traffic accordingly based
> on the method you choose; it is a pretty smart protocol. By default the
> distribution will be in relation to the metric, in other words it will be
> proportional. I think this link will help some (watch the word wrap):
>
>
> http://www.cisco.com/en/US/tech/tk827/tk831/technologies_tech_note09186a0080094806.shtml
>
> When using CEF and EIGRP or really any load ballancing with CEF, there are
> some caveats (of course ...) depending on the amount of destinations your
> traffic is going to. Here is a helpful link with describes this some and in
> particular discusses CEF and EIGRP:
>
>
> http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a008009437d.shtml
>
> HTH,
>
> Andrew Lee Lissitz
>
>
>
>
> On Wed, Apr 22, 2009 at 10:18 AM, Dominik Nowicki <freemaxis_at_gmail.com>wrote:
>
>> Hi GS,
>> I have doubts regarding EIGRP Unequal Cost Load Balancing.
>>
>> I have a questions:
>>
>> 1. If we have CEF enabled on both interfaces and just add variance 2
>> command
>> under EIGRP, then for example we achived traffic share count like 2:1, but
>> how really traffic is sending to the remote devices. With CEF on interface
>> and variance we have per-destination sharing load sharing.
>> Does it mean that for examle if we have just one route that appears in the
>> routing table with 2:1 ratio and we ping from our router (always with the
>> same source IP to the same destination) then we really dont achived any
>> load
>> balancing because CEF sends the packets based on the created flow (src and
>> dest IP, so always the same).
>> So what exactly this variance + CEF do for us? Does it mean that to take
>> adavantage from the variance we have to disable CEF?
>>
>> 2. If we add ip load-sharing per-packet command under interface then we
>> have per-packet sharing under show ip cef X.X.X.X, does it mean that
>> thanks
>> to this command we disabled CEF on this interfaces and we have packets
>> that
>> are process switched.
>>
>>
>> Thanks for clarigfications
>>
>>
>> Blogs and organic groups at http://www.ccie.net
>>
>> _______________________________________________________________________
>> Subscription information may be found at:
>> http://www.groupstudy.com/list/CCIELab.html
>>
>>
>>
>>
>>
>>
>>
>>
>
>
> --
> Andrew Lee Lissitz
> all.from.nj_at_gmail.com
>
-- Andrew Lee Lissitz all.from.nj_at_gmail.com Blogs and organic groups at http://www.ccie.netReceived on Wed Apr 22 2009 - 14:15:53 ART
This archive was generated by hypermail 2.2.0 : Mon May 04 2009 - 07:39:12 ART