[nsp] ospf default routes and bgp injection

[Mark Tinka]

Nick Kraal wrote:
> Mark,

Hi Nick.

> 
> The arguments presented are definitely valid. Scalability and
> convergence on top of the list. But some folks like to see
> documents/white papers from the vendors to be convinced. Well thanks
> you all for the input.   

Well, unoficially, I am going to claim that I haven't yet seen a document on
www.cisco.com that recommends this scenario (perhaps it's now there since
the last time I checked).

Officially, I will claim that one of Cisco's brightest engineers and key
personnel *highly* recommended the scenarion at a recent NOG in East Africa,
and he thinks it will scale best. Coming from him, I'd assume Cisco's
position is generally in sync :).

Sorry for the vagueness.

> 
> Regards,
> 
> -nick/
> 
> ----- Original Message -----
> From: "Mark Tinka" <mtinka at africaonline.co.sz>
> To: "'Oliver Boehmer (oboehmer)'" <oboehmer at cisco.com>; "'Nick
> Kraal'" <nick at arc.net.my>; <cisco-nsp at puck.nether.net> Sent:
> Wednesday, April 07, 2004 5:05 PM Subject: RE: [nsp] ospf default
> routes and bgp injection 
> 
> 
>> cisco-nsp-bounces at puck.nether.net wrote:
>>> Nick,
>>> 
>>> I think it really depends on what you want to achieve and wich IGP
>>> you're using. As general arguments:
>>> 
>>> 1) customer routes in IGP don't scale. While OSPF/ISIS on modern
>>> platforms (RISC CPU's) works just fine with 5000+ routes in your
>>> IGP, scaling to 10000 and above might be a challenge
>> 
>> As Nick mentioned, using OSPF to carry your infrastructure networks
>> (Loopback interface IP's mostly) and using iBGP to carry your
>> internal and customer prefixes is quite a scalable approach.
>> 
>> Nick (sorry to post this in Oliver's thread), you're right. I don't
>> think
> I
>> have seen a supporting document anywhere yet (there could be,
>> though), but this is something that has been presented and agreed on
>> multiple operators forums and seminars that I've been a part of.
>> 
>> It scales quite well and offers more features in terms of filtering,
>> flexibility and all. Just FYI, this is how I run my backbone.
>> 
>>> 
>>> 2) The more routes you carry in your IGP, the longer PRC takes
>>> installing the routes in your RIB. This will be in issue if you want
>>> to tune your IGP for faster (i.e. sub-second) convergence
>>> 
>>> I think one can find addtl. arguments if your specific network and
>>> routing topology is taken into account.
>>> 
>>> oli
>>> 
>>> ----Original Message----
>>> From: cisco-nsp-bounces at puck.nether.net
>>> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Nick Kraal
>>> Sent: Dienstag, 6. April 2004 14:45 To: cisco-nsp at puck.nether.net
>>> Subject: Re: [nsp] ospf default routes and bgp injection
>>> 
>>>> Following up on this one. We all know that the IGP of choice used
>>>> should carry only infrastructure networks and kept as small as
>>>> possible. The rest should be announced via iBGP. Is there a
>>>> document that explains this rationale clear --having some
>>>> difficulty convincing some folks.
>>>> 
>>>> Thanks in advance,
>>>> 
>>>> -nick/
>>>> 
>>>> ----- Original Message -----
>>>> From: "Gert Doering" <gert at greenie.muc.de>
>>>> To: "Christopher J. Wolff" <chris at bblabs.com>
>>>> Cc: <cisco-nsp at puck.nether.net>
>>>> Sent: Thursday, April 01, 2004 5:49 AM
>>>> Subject: Re: [nsp] ospf default routes and bgp injection
>>>> 
>>>> 
>>>>> Hi,
>>>>> 
>>>>> On Wed, Mar 31, 2004 at 09:58:12AM -0700, Christopher J. Wolff
>>>>> wrote:
>>>>>> I've considered ibgp as an end-to-end solution however never went
>>>>>> so far as to implement it.
>>>>> 
>>>>> Completely overkill in this network.  If there's no BGP speaker
>>>>> "down the road", there is not much use in having all those routers
>>>>> actually carry BGP information.
>>>>> 
>>>>>> My concern is making sure that the entire routing table isn't
>>>>>> propagated all the way to the edge device, which could be
>>>>>> something minimal like a 2611XM.  Any thoughts?  Thank you for
>>>>>> your advice.
>>>>> 
>>>>> Have the BGP speakers distribute an OSPF (or EIGRP, or even RIP :)
>>>>> ) default route to the smaller boxes.
>>>>> 
>>>>> gert
>>>>> --
>>>>> USENET is *not* the non-clickable part of WWW!
>>>>> 
>>>> //www.muc.de/~gert/
>>>>> Gert Doering - Munich, Germany gert at greenie.muc.de
>>>>> fax: +49-89-35655025
>>>> gert at net.informatik.tu-muenchen.de
>>>>> _______________________________________________
>>>>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>>>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>>>>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>>>> 
>>>> 
>>>> _______________________________________________
>>>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>>>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>> 
>>> _______________________________________________
>>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>> https://puck.nether.net/mailman/listinfo/cisco-nsp archive at
>>> http://puck.nether.net/pipermail/cisco-nsp/
>> 
>> Regards,
>> 
>> Mark Tinka
>> Technical Manager, Africa Online Swaziland

Regards,

Mark Tinka
Technical Manager, Africa Online Swaziland