RE: DLSW backup peer.

From: omar guarisco (oguarisco44@xxxxxxxxxxx)
Date: Tue Jan 08 2002 - 20:11:50 GMT-3


   
maybe using TCP the reconvergence time is less that using FST...

If you want to control your DLSW circuit then you can set R2 and R3 as
promiscuos or define the dlsw local peer as passive so ciircuit will be
established only from End-Point DLSw to DLSw "concentrator"

me too was implementing a solution using Backup peer...but I think i'll use
the cost defined per peer so that we have 2 DLSw circuit opened
- from R1 to R2 cost 4
- from R1 to R3 cost 2

simultaneously...knowing that the circuit used would ever be the second one
with less cost...maybe this will increase the convergence timer more if it's
use in addition to TCP encapsulation

R1
> >dlsw local-peer peer-id 137.6.2.2
> >dlsw remote-peer 0 tcp 137.6.5.5 cost 4
> >dlsw remote-peer 0 tcp 137.6.3.3 cost 2

TO test !!

Obviously this depend on your DLSW designing, peer relationship and the
routers...

If you have 100 remote peer DLSW that connect to 2 DLSW concetrator in a
hub&spoke topology then with backup you're sure that only 100 circuit will
be opened (maybe using load-sharing 50 on R2 and 50 on R3) but using backup
with cost then 100 circuit will be opened on R2 and 100 on R3...

If you want to add more complexity you can configure enhanced Load-Balancing
using Circuit weight....

Omar

>From: "McCallum, Robert" <Robert.McCallum@let-it-be-thus.com>
>Reply-To: "McCallum, Robert" <Robert.McCallum@let-it-be-thus.com>
>To: "'Stephen Oliver'" <stevie_oliver@hotmail.com>, ccielab@groupstudy.com
>Subject: RE: DLSW backup peer.
>Date: Tue, 8 Jan 2002 17:28:38 -0000
>
>actually no looking at your configs why use fst. try using tcp see if
>that's any faster.
>Also when using backup peers in such a manner you should be using
>promiscuous on R2 and R3 with no remote-peer statements. I know that if
>you use remote-peers when using TCP the peering gets into a whole lot of
>mess.
>
>-----Original Message-----
>From: Stephen Oliver [mailto:stevie_oliver@hotmail.com]
>Sent: 08 January 2002 17:21
>To: Robert.McCallum@let-it-be-thus.com; ccielab@groupstudy.com
>Subject: RE: DLSW backup peer.
>
>
>thanks for the hint, hint but....
>
>I know I can use the linger keyword to hold the backup peer for a length of
>time after the primary peer comes back.. That's not what I asked though.
>
>
> >From: "McCallum, Robert" <Robert.McCallum@let-it-be-thus.com>
> >To: 'Stephen Oliver' <stevie_oliver@hotmail.com>, ccielab@groupstudy.com
> >Subject: RE: DLSW backup peer.
> >Date: Tue, 8 Jan 2002 17:08:45 -0000
> >
> >try lingering about.....hint hint
> >
> >-----Original Message-----
> >From: Stephen Oliver [mailto:stevie_oliver@hotmail.com]
> >Sent: 08 January 2002 16:56
> >To: ccielab@groupstudy.com
> >Subject: DLSW backup peer.
> >
> >
> >How long should it take for a peer to go into disconnect and a backup to
> >connect in a DLSW connection. I have one router R1 with R2 as its
>primary
> >peer and R3 as its backup peer. When I kill the link to the primary peer
> >it
> >can take 1-2 mins to switch over to the backup peer and the same to
>switch
> >back when the primary peer is reachable again. Is this normal. Is there
> >any
> >way to speed it up. The relevant parts of the configs are below.
> >
> >Thanks, Stephen.
> >
> >R1
> >dlsw local-peer peer-id 137.6.2.2
> >dlsw remote-peer 0 fst 137.6.5.5
> >dlsw remote-peer 0 fst 137.6.3.3 backup-peer 137.6.5.5
> >dlsw circuit-keepalives
> >
> >R2
> >dlsw local-peer peer-id 137.6.5.5
> >dlsw remote-peer 0 fst 137.6.2.2
> >dlsw circuit-keepalives
> >
> >R3
> >dlsw local-peer peer-id 137.6.3.3
> >dlsw remote-peer 0 fst 137.6.2.2
> >dlsw circuit-keepalives
> >
> >
> >
> >
> >
> >



This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:56:20 GMT-3