alt.certification.cisco

Author

BGP Multihop

Ian Blaney

2003-01-24, 5:24 pm

Hi Folks

My head's sore thinking about this. Am I missing something obvious
here. I have the following setup.

AS1 AS2
BGP BGP
R1---------R2-------------------R3---------R4

R1 is peering with R4. R1 is not directly connected to R4 buts goes
via R2 and R3 which do not run bgp as they are too small to hold the
full bgp routing table. I guess to get the peering established I have
to configure Multihop and have a tcp connection from R1 to R4.

My question is how do I configure R2 and R3. This is where the head
starts hurting. If I configure a default route on R2 to point to R3
and another default route on R3 to R4 how are the packets going to get
back to R1 from R4. I will have a routing loop when R4 sends the
packet to R3 and then R3 sends it back to R4 via the default gateway.

Am I missing something on how bgp works as I don't have much practical
experience with it or do I have to do some form of redistribution.

Could someone please give an idea and if possible a short config file
for this solution.

Thanks

Hansang Bae

2003-01-24, 9:24 pm

In article <3e31c82a.33816777@news.easynews.com>, Isblaney@aol.comXX
says...
> Hi Folks
>
> My head's sore thinking about this. Am I missing something obvious
> here. I have the following setup.
>
> AS1 AS2
> BGP BGP
> R1---------R2-------------------R3---------R4
>
> R1 is peering with R4. R1 is not directly connected to R4 buts goes
> via R2 and R3 which do not run bgp as they are too small to hold the
> full bgp routing table. I guess to get the peering established I have
> to configure Multihop and have a tcp connection from R1 to R4.
> My question is how do I configure R2 and R3. This is where the head
> starts hurting. If I configure a default route on R2 to point to R3
> and another default route on R3 to R4 how are the packets going to get
> back to R1 from R4. I will have a routing loop when R4 sends the
> packet to R3 and then R3 sends it back to R4 via the default gateway.
> Am I missing something on how bgp works as I don't have much practical
> experience with it or do I have to do some form of redistribution.
> Could someone please give an idea and if possible a short config file
> for this solution.

This is one of the dangers of multiphop. Basically, you need to let R2
and R3 know how to reach the IP addresses that you're going to use for
R1 and R4's BGP. For that matter, R4 needs to know that its neighbor's
(R1) IP can be reached via R3. And R1 needs to know that its neighbor
(R4) can be reached via R2. Static routes will do fine.

--

hsb

"Somehow I imagined this experience would be more rewarding" Calvin
*************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
******************************
******************************
********
Due to the volume of email that I receive, I may not not be able to
reply to emails sent to my account. Please post a followup instead.
******************************
******************************
********

Tom Gill

2003-01-25, 1:23 am

I concur.

Tom
CCNP

"Hansang Bae" <uonr@alp.ee.pbz> wrote in message
news:MPG.189bbf7955e581ac989753@news-server.nyc.rr.com...
> In article <3e31c82a.33816777@news.easynews.com>, Isblaney@aol.comXX
> says...
> > Hi Folks
> >
> > My head's sore thinking about this. Am I missing something obvious
> > here. I have the following setup.
> >
> > AS1 AS2
> > BGP BGP
> > R1---------R2-------------------R3---------R4
> >
> > R1 is peering with R4. R1 is not directly connected to R4 buts goes
> > via R2 and R3 which do not run bgp as they are too small to hold the
> > full bgp routing table. I guess to get the peering established I have
> > to configure Multihop and have a tcp connection from R1 to R4.
> > My question is how do I configure R2 and R3. This is where the head
> > starts hurting. If I configure a default route on R2 to point to R3
> > and another default route on R3 to R4 how are the packets going to get
> > back to R1 from R4. I will have a routing loop when R4 sends the
> > packet to R3 and then R3 sends it back to R4 via the default gateway.
> > Am I missing something on how bgp works as I don't have much practical
> > experience with it or do I have to do some form of redistribution.
> > Could someone please give an idea and if possible a short config file
> > for this solution.
>
>
> This is one of the dangers of multiphop. Basically, you need to let R2
> and R3 know how to reach the IP addresses that you're going to use for
> R1 and R4's BGP. For that matter, R4 needs to know that its neighbor's
> (R1) IP can be reached via R3. And R1 needs to know that its neighbor
> (R4) can be reached via R2. Static routes will do fine.
>
> --
>
> hsb
>
> "Somehow I imagined this experience would be more rewarding" Calvin
> *************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
> ******************************
******************************
********
> Due to the volume of email that I receive, I may not not be able to
> reply to emails sent to my account. Please post a followup instead.
> ******************************
******************************
********
>

Ian Blaney

2003-01-25, 10:23 am

Hi Hansang

Thanks for your reply but I am still confused (Not very difficult
these days). Can I draw a diagram with IPs to make it clear in my
mind.

AS1 AS2
1.1.1.1 5.5.5.5
BGP BGP
2.2.2.0/30 3.3.3.0/30 4.4.4.0/30

R1---------------------R2--------------------R3--------------------R4
.1 .2 .1 .2 .1 .2

R1 has a loopback address 1.1.1.1 used for peering.
R4 " " " 5.5.5.5 " " "

If I set the following

R1 to get to R4 (5.5.5.5) next hop 2.2.2.2
R2 to get to R4 (5.5.5.5) next hop 3.3.3.2
R3 to get to R4 (5.5.5.5) next hop 4.4.4.2

R4 to get to R1 (1.1.1.1) next hop 4.4.4.1
R3 to get to R1 (1.1.1.1) next hop 3.3.3.1
R2 to get to R1 (1.1.1.1) next hop 2.2.2.1

If I setup the static routes described above I agree the BGP peering
will work but for talking sake, say a user in AS1 wants to reach a web
server on the internet. If AS1's normal ISP is down or the BGP path
selection chooses to go via AS2. How does the packet know where to go
when it reaches R2 as it doesn't have the full BGP routing table.
Likewise with R3. If I set a default route on R2 pointing to R3 and a
default route on R3 to R4 the packet will reach AS2 but what happens
if I have this setup and AS2 goes down and a user in AS2 wants to
reach the webserver via AS1. When the packets reach R3 they will be
sent out the default gateway, as R3 does not have the full routing
table, back to R4. We will have a routing loop.

Sorry if I didn't make myself clear before but AS1 and AS2 have their
own ISP connections. The eBGP connection between them is used for
redundancy in case either goes down.

Am I missing something obvious here???

Regards
Ian Blaney

On Sat, 25 Jan 2003 02:22:20 GMT, Hansang Bae <uonr@alp.ee.pbz> wrote:

>In article <3e31c82a.33816777@news.easynews.com>, Isblaney@aol.comXX
>says...
>> Hi Folks
>>
>> My head's sore thinking about this. Am I missing something obvious
>> here. I have the following setup.
>>
>> AS1 AS2
>> BGP BGP
>> R1---------R2-------------------R3---------R4
>>
>> R1 is peering with R4. R1 is not directly connected to R4 buts goes
>> via R2 and R3 which do not run bgp as they are too small to hold the
>> full bgp routing table. I guess to get the peering established I have
>> to configure Multihop and have a tcp connection from R1 to R4.
>> My question is how do I configure R2 and R3. This is where the head
>> starts hurting. If I configure a default route on R2 to point to R3
>> and another default route on R3 to R4 how are the packets going to get
>> back to R1 from R4. I will have a routing loop when R4 sends the
>> packet to R3 and then R3 sends it back to R4 via the default gateway.
>> Am I missing something on how bgp works as I don't have much practical
>> experience with it or do I have to do some form of redistribution.
>> Could someone please give an idea and if possible a short config file
>> for this solution.
>
>
>This is one of the dangers of multiphop. Basically, you need to let R2
>and R3 know how to reach the IP addresses that you're going to use for
>R1 and R4's BGP. For that matter, R4 needs to know that its neighbor's
>(R1) IP can be reached via R3. And R1 needs to know that its neighbor
>(R4) can be reached via R2. Static routes will do fine.
>
>--
>
>hsb
>
>"Somehow I imagined this experience would be more rewarding" Calvin
>*************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
> ******************************
******************************
********
>Due to the volume of email that I receive, I may not not be able to
>reply to emails sent to my account. Please post a followup instead.
> ******************************
******************************
********

Thomas Kuborn

2003-01-27, 4:23 am

Dear,

What exactly are you trying to accomplish ?
Are you AS1 & your unique ISP is AS2 ?

Thomas

"Ian Blaney" <Isblaney@aol.comXX> wrote in message
news:3e32b0c2.19505303@news.easynews.com...
> Hi Hansang
>
> Thanks for your reply but I am still confused (Not very difficult
> these days). Can I draw a diagram with IPs to make it clear in my
> mind.
>
> AS1 AS2
> 1.1.1.1 5.5.5.5
> BGP BGP
> 2.2.2.0/30 3.3.3.0/30 4.4.4.0/30
>
> R1---------------------R2--------------------R3--------------------R4
> .1 .2 .1 .2 .1 .2
>
> R1 has a loopback address 1.1.1.1 used for peering.
> R4 " " " 5.5.5.5 " " "
>
> If I set the following
>
> R1 to get to R4 (5.5.5.5) next hop 2.2.2.2
> R2 to get to R4 (5.5.5.5) next hop 3.3.3.2
> R3 to get to R4 (5.5.5.5) next hop 4.4.4.2
>
> R4 to get to R1 (1.1.1.1) next hop 4.4.4.1
> R3 to get to R1 (1.1.1.1) next hop 3.3.3.1
> R2 to get to R1 (1.1.1.1) next hop 2.2.2.1
>
> If I setup the static routes described above I agree the BGP peering
> will work but for talking sake, say a user in AS1 wants to reach a web
> server on the internet. If AS1's normal ISP is down or the BGP path
> selection chooses to go via AS2. How does the packet know where to go
> when it reaches R2 as it doesn't have the full BGP routing table.
> Likewise with R3. If I set a default route on R2 pointing to R3 and a
> default route on R3 to R4 the packet will reach AS2 but what happens
> if I have this setup and AS2 goes down and a user in AS2 wants to
> reach the webserver via AS1. When the packets reach R3 they will be
> sent out the default gateway, as R3 does not have the full routing
> table, back to R4. We will have a routing loop.
>
> Sorry if I didn't make myself clear before but AS1 and AS2 have their
> own ISP connections. The eBGP connection between them is used for
> redundancy in case either goes down.
>
> Am I missing something obvious here???
>
> Regards
> Ian Blaney
>
> On Sat, 25 Jan 2003 02:22:20 GMT, Hansang Bae <uonr@alp.ee.pbz> wrote:
>
> >In article <3e31c82a.33816777@news.easynews.com>, Isblaney@aol.comXX
> >says...
> >> Hi Folks
> >>
> >> My head's sore thinking about this. Am I missing something obvious
> >> here. I have the following setup.
> >>
> >> AS1 AS2
> >> BGP BGP
> >> R1---------R2-------------------R3---------R4
> >>
> >> R1 is peering with R4. R1 is not directly connected to R4 buts goes
> >> via R2 and R3 which do not run bgp as they are too small to hold the
> >> full bgp routing table. I guess to get the peering established I have
> >> to configure Multihop and have a tcp connection from R1 to R4.
> >> My question is how do I configure R2 and R3. This is where the head
> >> starts hurting. If I configure a default route on R2 to point to R3
> >> and another default route on R3 to R4 how are the packets going to get
> >> back to R1 from R4. I will have a routing loop when R4 sends the
> >> packet to R3 and then R3 sends it back to R4 via the default gateway.
> >> Am I missing something on how bgp works as I don't have much practical
> >> experience with it or do I have to do some form of redistribution.
> >> Could someone please give an idea and if possible a short config file
> >> for this solution.
> >
> >
> >This is one of the dangers of multiphop. Basically, you need to let R2
> >and R3 know how to reach the IP addresses that you're going to use for
> >R1 and R4's BGP. For that matter, R4 needs to know that its neighbor's
> >(R1) IP can be reached via R3. And R1 needs to know that its neighbor
> >(R4) can be reached via R2. Static routes will do fine.
> >
> >--
> >
> >hsb
> >
> >"Somehow I imagined this experience would be more rewarding" Calvin
> >*************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
> > ******************************
******************************
********
> >Due to the volume of email that I receive, I may not not be able to
> >reply to emails sent to my account. Please post a followup instead.
> > ******************************
******************************
********
>

Barry Margolin

2003-01-27, 10:24 am

In article <3e32b0c2.19505303@news.easynews.com>,
Ian Blaney <Isblaney@aol.comXX> wrote:
>If I setup the static routes described above I agree the BGP peering
>will work but for talking sake, say a user in AS1 wants to reach a web
>server on the internet. If AS1's normal ISP is down or the BGP path
>selection chooses to go via AS2. How does the packet know where to go
>when it reaches R2 as it doesn't have the full BGP routing table.
>Likewise with R3. If I set a default route on R2 pointing to R3 and a
>default route on R3 to R4 the packet will reach AS2 but what happens
>if I have this setup and AS2 goes down and a user in AS2 wants to
>reach the webserver via AS1. When the packets reach R3 they will be
>sent out the default gateway, as R3 does not have the full routing
>table, back to R4. We will have a routing loop.
>
>Sorry if I didn't make myself clear before but AS1 and AS2 have their
>own ISP connections. The eBGP connection between them is used for
>redundancy in case either goes down.

If you're just trying to get to the Internet, you don't need full BGP
tables. You can have the ISPs just send you default routes. In this case,
R2 and R3 should be able to run BGP, and you won't need to use multi-hop.

--
Barry Margolin, barmar@genuity.net
Genuity, Woburn, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.

Sponsored Links