[c-nsp] TE config help

[mb2 at os.datafx.com.au]

Quoting Pete Templin <petelists at templin.org>:

> mb2 at os.datafx.com.au wrote:
>> interface FastEthernet2/0
>> ### Would this still be required? ip ospf cost 100
>
> I'd definitely recommend a sane IGP design, so that if TE should 
> explode you'll fail back to predictable IGP routing.

Ok - Ta.

>
>> ### Not too sure about this? ip rsvp bandwidth ? ?
>
> ip rsvp band 4000 4000
>
>> interface ATM1/0.100 point-to-point
>> ### Not too sure about this? ip rsvp bandwidth ? ?
>
> ip rsvp band 8000 8000
>
>> interface Tunnel2  tag-switching ip
>
> Are you carrying MPLS VPNs across these links, AND these routers are 
> not the ultimate hop?  If not, why do you have 'mpls ip' on these 
> tunnels?

Yes - We have mpls vpns running over these links...the 7200's are 
performing edge+core functionality....so I assume we will require "mpls 
ip" on the tunnels?


>
>> ### Not too sure about this? tunnel mpls traffic-eng bandwidth ?
>
> tunn mpls traf band 4000

Ok - Little confused here....why wouldn't this be:

tunn mpls traf band 12000?

Why would I have 3 tunnels @ 4Mb each?

>
>> tunnel mpls traffic-eng path-option 5 dynamic
>
> How many tunnels are you going to have?  You can do three tunnels, 
> each at 4Mbps bandwidth.  You can do two or three or more with 
> auto-bw.

What determines/dictates which method should be used?

>
> Your biggest detail is likely going to be how well your traffic 
> splits up into parallel tunnels.  If it hashes out well, two or three 
> static tunnels will fit right into your interfaces.  If it ends up 
> quite dynamic, auto-bw might be a real good answer for you.  If it 
> doesn't hash well and stays that way, you may simply have your hands 
> full experimenting.

Thanks very much for your suggestions + explanations - Really appreciate it.

>
> pt
>



-------------------------------------------------------------------------
This e-mail was sent via Data FX Online WebMail http://www.datafx.com.au/