[MLB-WIRELESS] OSPF BGP

[Dan Flett]

I don't think there needs to be any sort of official "decision" to *test*
BGP.  We're not scrapping OSPF - any testing I do will fully support anyone
who uses OSPF.

At the moment we're not even using OSPF the way the Melbourne Wireless
guidelines say we should.  Everyone I know of is using OSPF Area 0, because
experience shows that's the only way that properly works in our situation.
When we use OSPF Area numbers, routing breaks because we don't connect to
each other the way OSPF expects us to.  And we shouldn't have to.  We should
have a routing protocol that works for us the way we operate naturally.  We
shouldn't have to plan our links to fit the routing protocol.

If we waited for official decisions to test new technologies and protocols
we'd progress a lot slower than we already do.  If someone wants to try
something out, they should do so - so long as it doesn't break the network
for everybody else.  When the technology under test turns out to be a better
alternative than the current official standard, that's the time to weigh the
pros and cons and make a decision.

Dan

-----Original Message-----
From: owner-melbwireless at wireless.org.au
[mailto:owner-melbwireless at wireless.org.au] On Behalf Of Ryan Abbenhuys
Sent: Wednesday, 22 June 2005 3:04 PM
To: melbwireless at wireless.org.au
Subject: Re: [MLB-WIRELESS] OSPF BGP


Did I miss something?  Was the decision to use OSPF scrapped in favour of
BGP?


>http://www.melbourne.wireless.org.au/wiki/?BGP
>
>
>ivile01 at yahoo.com.au | ivile at ivile.bur.st http://bur.st/~ivile 
>(waveguides) | http://ivile.bur.st | http://ivile.bur.st/ivile/64/ (my 
>car) http://www.melbourne.wireless.org.au/users/?ivile
>----- Original Message -----
>From: "Brenton D." <ivile01 at yahoo.com.au>
>To: "Dan Flett" <conhoolio at hotmail.com>; "'Nigel'" 
><thenigel at hotmail.com>;

><melbwireless at wireless.org.au>
>Sent: Wednesday, June 22, 2005 2:10 PM
>Subject: Re: [MLB-WIRELESS] OSPF BGP
>
>
>> Ill put up a really basic sample config. On the Wiki And ill have a 
>> part where you can add your own asn.
>> I was thinking that we should have a 5 number gap between each person
BGP 
>> asn, just in case they have more than one router(like me).
>> so basically we have room for 1000 routers 64512 to 65534 which i 
>> doubt there will be more than 50 running bgp in the near future.(as 
>> some only support ospf)
>>
>> so node fut would have the bgp asn 64515 (just leave the first few free)
>>            fuu would have the bgp asn 64520
>>            gho would have the bgp asn 64525 and so on...  unless you 
>> have on the node page that they request a BGP
>> asn(s)  from melb-wireless.
>>
>> ivile01 at yahoo.com.au | ivile at ivile.bur.st http://bur.st/~ivile 
>> (waveguides) | http://ivile.bur.st | http://ivile.bur.st/ivile/64/ 
>> (my car) http://www.melbourne.wireless.org.au/users/?ivile
>> ----- Original Message -----
>> From: "Dan Flett" <conhoolio at hotmail.com>
>> To: "'Brenton D.'" <ivile01 at yahoo.com.au>; "'Nigel'" 
>> <thenigel at hotmail.com>; <melbwireless at wireless.org.au>
>> Sent: Wednesday, June 22, 2005 1:48 PM
>> Subject: RE: [MLB-WIRELESS] OSPF BGP
>>
>>
>>> So, Brenton, cutting it right down, your BGP file looks like this:
>>>
>>> !
>>> hostname bgpd
>>> password XXXXX
>>> enable password XXXXX
>>> !
>>> router bgp 7675
>>> bgp router-id 10.10.129.145
>>> redistribute ospf
>>> redistribute connected
>>> !
>>> ! DAN's COMMENTS: you probably don't need redistribute connected as 
>>> you've
>>> already declared your network
>>> ! In Quagga the routing protocol automatically redistributes any routes
>>> declared with the network statement
>>> ! Also, to be precise, the melbourne wireless network is entirely
inside 
>>> the
>>> 10.10.0.0/16 supernet
>>> !
>>> network 10.0.0.0/8
>>> neighbor 10.10.128.97 remote-as 7676
>>> !
>>> access-list all permit any
>>> ! You probably don't need this access-list because you haven't
specified 
>>> any
>>> route-maps
>>> log stdout
>>>
>>>
>>> That's as simple as a BGP file gets really - and if we weren't using
OSPF
>>> you could get rid of the redistribute ospf statement too.  You only
need 
>>> to
>>> add neighbor lines each time you directly connect to a new BGP
neighbor.
>>> BGP gets complicated when you have multiple routes/routers/subnets
within
>>> the one AS.  If every node has their own AS it's quite easy.
>>>
>>> I'm considering writing a set of scripts that will automagically create
>>> Quagga/BGP config files from NVRAM variables or a very basic config
file,
>>> and that will exchange AS information with neighbors via DHCP.  So 
>>> basically
>>> you won't have to do anything (if you don't want to) except enter your
IP
>>> and AS addresses/numbers to start with.
>>>
>>> I'm thinking of running BGP at node GHO alongside OSPF.  We should let 
>>> GHO
>>> settle and make sure it's stable for a few weeks before we try
anything, 
>>> but
>>> I think it would be worth testing.
>>>
>>> How about someone create a BGP-Trial wiki page where we write down our
AS
>>> numbers for our nodes?  It should just be for testing, but it means we 
>>> can
>>> test BGP in our own local clusters.  It doesn't matter if ASNs within a
>>> cluster are contiguous or not - that's the beauty of BGP - you can
choose
>>> any number you want, so long as it isn't someone else's.  We need to 
>>> choose
>>> our numbers from the IANA Private Use ASN space - being 64512 to 65534,
>>> inclusive.
>>>
>>> Cheers,
>>>
>>> Dan
>>
>>
>> To unsubscribe: send mail to majordomo at wireless.org.au
>> with "unsubscribe melbwireless" in the body of the message
>> 
>
>
>To unsubscribe: send mail to majordomo at wireless.org.au
>with "unsubscribe melbwireless" in the body of the message
>
>

To unsubscribe: send mail to majordomo at wireless.org.au
with "unsubscribe melbwireless" in the body of the message


To unsubscribe: send mail to majordomo at wireless.org.au
with "unsubscribe melbwireless" in the body of the message