Re: Redistribution Problem

I was always under the impression that it was first in would stay in (until withdrawn) when it came two routing processes with the same AD. But at least this way it gives you a predicatable behaviour you can work with.

If the man with the expensive tie really wanted it 3 ways you could solve it would be:

Split the network to be load balanced in half. Then on R5 set the AD for one 1/2 down in one AS and the other down in the other. Not true load balancing and it wouldn't work with /32.

2x GRE tunnels. One via R3-R5 and the other via R4-R5 then extend out the OSPF domain to R5. Then tweak the metrics to get the network to load balance on R5.

MPLS TE. set up 2 paths one via R5 - R3 - R2 - R1 and the other R5 - R4 - R1

Are there any others that I have missed (or an easier one)?

Question for the group, when you have 2 OSPF routing processes with the same prefix is it the lowest process ID that gets chosen? Just wondering if this is common across the protocols now.

Vince

On 28/08/2010 8:28 AM, Kambiz Agahian wrote:
> Hemanth,
>
> Now the million dollar question is how to enable load balancing in scenarios
> like this ? :) ... we can think of several ways.
>
> PS. Just to make sure everyone's clear on the last point and everybody's on
> the same page, in this case since we're learning the routes from two
> different AS's (AS - Autonomous System - just think about
> it...."autonomous" what does it mean?) both ways of route selection (prior
> to 12.2 and beyond that) make perfect sense. And the LB option is not such a
> great idea.
>
> In a normal situation however you probably don't want to load balance across
> two different networks (i.e. different management, routing policies, fees,
> SLA's etc.).
>
> Older versions of IOS (12.2 and again some switches) take the path that they
> hear about first. On the other hand, more recent versions of IOS choose the
> path with the lowest AS number.* Just a sort of tie breaker*.
>
> But what if your boss the guy with that expensive tie and belt keeps saying
> "load balancing"?! I always believe when something obvious (even in IOS 15)
> is not there probably it should not be! but how would you save your team?
>
>
>
> Kambiz Agahian
> CCIE Instructor/Consultant
> M.Eng Telecom, CCIE# 25341, CCSI# 33326, MCSE, MCSA
>
>
> On Fri, Aug 27, 2010 at 10:42 AM, Kambiz Agahian<aussiecert_at_gmail.com>wrote:
>
>> Yeah as the otehr user posted the link and I asked before, you will see
>> either the route picked based on timestamp OR based on AS numbers. With the
>> IOS version you're using (12.4T) the one with the lower AS number takes
>> precedence. But again, this is still the case with old versions of IOS and I
>> believe still some switches.
>>
>>
>> HTH
>>
>>
>> Kambiz Agahian
>> CCIE Instructor/Consultant
>> M.Eng Telecom, CCIE# 25341, CCSI# 33326, MCSE, MCSA
>>
>>
>> On Fri, Aug 27, 2010 at 9:49 AM, HEMANTH RAJ<hemanthrj_at_gmail.com> wrote:
>>
>>> No wat i have done first with AS 100 and then AS 200. It goes thorugh AS
>>> 100
>>> And then created another interface with another in AS 50 ,then afterwards
>>> it is goin via AS 50
>>> tats the confusing point!!!
>>>
>>>
>>> On Fri, Aug 27, 2010 at 10:17 PM, Kambiz Agahian<aussiecert_at_gmail.com>wrote:
>>>
>>>> The one with a lower number or the one to be seen first?
>>>>
>>>> Kambiz Agahian
>>>> CCIE Instructor/Consultant
>>>> M.Eng Telecom, CCIE# 25341, CCSI# 33326, MCSE, MCSA
>>>>
>>>> On Fri, Aug 27, 2010 at 9:43 AM, HEMANTH RAJ<hemanthrj_at_gmail.com>wrote:
>>>>
>>>>> yes Vince u r right it chooses the one with a lower Autonomous system
>>>>> .Experts pls guide me for this question.
>>>>>
>>>>>
>>>>>
>>>>> On Fri, Aug 27, 2010 at 11:35 AM, Vince Librandi<
>>>>> vlibrandi_at_internode.on.net
>>>>>> wrote:
>>>>>> Woops forgot to add the answer to your question.
>>>>>>
>>>>>> The router will only choose use the routes for a prefix from one
>>>>> routing
>>>>>> protocol because that way it can guarantee at least for that AS it has
>>>>> a
>>>>>> loop free path for the prefix (of course redistribution can negate
>>>>> that).
>>>>>> But if the router where to install one route from one AS and another
>>>>> from a
>>>>>> second AS it wouldn't be able to be sure that the path doesn't loop
>>>>>> somewhere as the two AS's don't share routing information.
>>>>>>
>>>>>> So it looks like the router chooses the lowest AS number when the
>>>>> prefix
>>>>>> length and AD are the same for a prefix.
>>>>>>
>>>>>>
>>>>>> On 27/08/2010 3:05 PM, Vince Librandi wrote:
>>>>>>
>>>>>>> Hi Hemanth,
>>>>>>>
>>>>>>> I just lab'ed up your set up and got the same affect. Looks like when
>>>>> you
>>>>>>> have two EIGRP process and the metric is the same and the AD on the
>>>>> route is
>>>>>>> the same the router chooses the process with the lowest AS number.
>>>>> Just to
>>>>>>> prove it I configured another EIGRP process (AS 50) on the R5-R4 link
>>>>> and
>>>>>>> the router choose that path instead.
>>>>>>>
>>>>>>> R5(config-router)#do sh ip ei top
>>>>>>> IP-EIGRP Topology Table for AS(100)/ID(10.1.45.5)
>>>>>>>
>>>>>>> Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,
>>>>>>> r - reply Status, s - sia Status
>>>>>>>
>>>>>>> P 1.1.1.1/32, 0 successors, FD is Inaccessible
>>>>>>> via 10.1.35.3 (3097600/2585600), Serial0/1
>>>>>>>
>>>>>>> IP-EIGRP Topology Table for AS(200)/ID(1.1.1.2)
>>>>>>>
>>>>>>> Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,
>>>>>>> r - reply Status, s - sia Status
>>>>>>>
>>>>>>> P 1.1.1.1/32, 0 successors, FD is Inaccessible
>>>>>>> via 10.1.45.4 (3097600/2585600), Serial0/0
>>>>>>>
>>>>>>> IP-EIGRP Topology Table for AS(50)/ID(10.1.45.5)
>>>>>>>
>>>>>>> Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,
>>>>>>> r - reply Status, s - sia Status
>>>>>>>
>>>>>>> P 1.1.1.1/32, 1 successors, FD is 3097600
>>>>>>> via 10.1.45.4 (3097600/2585600), Serial0/0
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> but if both links are in the same AS then I get the load balancing
>>>>>>>
>>>>>>> R5(config-router)#do sh ip ei top
>>>>>>> IP-EIGRP Topology Table for AS(100)/ID(10.1.45.5)
>>>>>>>
>>>>>>> Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,
>>>>>>> r - reply Status, s - sia Status
>>>>>>>
>>>>>>> P 1.1.1.1/32, 2 successors, FD is 3097600
>>>>>>> via 10.1.35.3 (3097600/2585600), Serial0/1
>>>>>>> via 10.1.45.4 (3097600/2585600), Serial0/0
>>>>>>>
>>>>>>> Looks like this is the way the routing table chooses between routes
>>>>> handed
>>>>>>> to it from different processes.
>>>>>>>
>>>>>>> Must see if OSPF does the same thing.
>>>>>>>
>>>>>>> Regards
>>>>>>>
>>>>>>> Vince
>>>>>>>
>>>>>>>
>>>>>>> On 27/08/2010 1:21 PM, HEMANTH RAJ wrote:
>>>>>>>
>>>>>>>> s0/1 s0/0
>>>>>>>> R2-------------------R3\s 0/2
>>>>>>>> | | \
>>>>>>>> |s0/0 s0/3 | \
>>>>>>>> | | \ R1-loo 0
>>>>>>>> 1.1.1.1/24
>>>>>>>> | | \ s 0/0 R1-loo 1
>>>>>>>> 11.11.11.11/24
>>>>>>>> | | \
>>>>>>>> | | / R5 R2-loo o
>>>>> 2.2.2.2
>>>>>>>> /24
>>>>>>>> | | / R2-loo 1
>>>>>>>> 22.22.22.22/24
>>>>>>>> | | / s 0/1
>>>>>>>> | | /
>>>>>>>> | s0/0 | /
>>>>>>>> | s0/3 | /s0/2
>>>>>>>> R1-------------------R4
>>>>>>>> s0/1 s0/0
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> I'm running ospf area 0 on R1,R2,R3 s 0/0,R4 s0/0
>>>>>>>>
>>>>>>>> R3 s0/3 ospf area 20
>>>>>>>> R4 s0/3 ospf area 20
>>>>>>>>
>>>>>>>> R3 s0/2 eigrp AS 100
>>>>>>>> R5 s0/0 eigrp AS 100
>>>>>>>>
>>>>>>>> R4 s0/2 eigrp AS 200
>>>>>>>> R5 s0/1 eigrp AS 200
>>>>>>>>
>>>>>>>>
>>>>>>>> I have done mutual redistribution on R3 between OSPF and EIGRP 100
>>>>>>>>
>>>>>>>> I have done mutual redistribution on R4 between OSPF and EIGRP 200
>>>>>>>>
>>>>>>>> Every route can ping every other route
>>>>>>>>
>>>>>>>> But when i check on R5 routing table I'm learning R1's loopback0 via
>>>>> R3
>>>>>>>> But as per the logic it should load balance between R3 and R4 but it
>>>>> is
>>>>>>>> choosing only R3
>>>>>>>> and when i check on R5 topology table it shows the route via R4 is
>>>>>>>> inaccessible
>>>>>>>>
>>>>>>>> I have reidstributed from opsf to eigrp with same metric on both R3
>>>>> and
>>>>>>>> R4
>>>>>>>>
>>>>>>>> R3
>>>>>>>> router eigrp 100
>>>>>>>> redis ospf 1 met 1 1 1 1 1
>>>>>>>>
>>>>>>>> R4
>>>>>>>> router eigrp 200
>>>>>>>> redis ospf 1 met 1 1 1 1 1
>>>>>>>>
>>>>>>>> But still R5 is choosing path via R3 and not loadbalancing between
>>>>> R3 and
>>>>>>>> R4
>>>>>>>>
>>>>>>>> Can anyone help me out of this issue???
>>>>>>>>
>>>>>>>
>>>>>>> Blogs and organic groups at http://www.ccie.net
>>>>>>>
>>>>>>>
>>>>> _______________________________________________________________________
>>>>>>> Subscription information may be found at:
>>>>>>> http://www.groupstudy.com/list/CCIELab.html
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>
>>>>> --
>>>>> Problems arise Bcoz we talk,prblms r not solve bcoz we dont talk So gud
>>>>> r
>>>>> bad talk to ur affectionate one's freely
>>>>>
>>>>> Urs Friendly,
>>>>> HP HEMANTH RAJ
>>>>>
>>>>>
>>>>> Blogs and organic groups at http://www.ccie.net
>>>>>
>>>>> _______________________________________________________________________
>>>>> Subscription information may be found at:
>>>>> http://www.groupstudy.com/list/CCIELab.html
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>
>>> --
>>> Problems arise Bcoz we talk,prblms r not solve bcoz we dont talk So gud r
>>> bad talk to ur affectionate one's freely
>>>
>>> Urs Friendly,
>>> HP HEMANTH RAJ
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html

Blogs and organic groups at http://www.ccie.net
Received on Sat Aug 28 2010 - 17:25:53 ART