Scenerio:- GRE Tunnel configured from points "A" to point "E" and works fine.
Between "A-B" "B-C" "C-D" "D-E" there are no QOS mechanism's. Links vary from 10mb/100mb/Gigabit.
If on the GRE Tunnel you place traffic shaping of say 5mb, does this then mean that "end to end" on all links 5mb is reservered for the GRE Tunnel across all links.
How is this policed/managed or controlled ?
Do you require the traffic shaping on both ends for this to work or is one end sufficient ?
Yes, that's correct, another reason to use a shaper, or policier, might be to keep some customer traffic within a contracted amount.
As to the question of insuring 5 Mbps for the tunnel traffic end-to-end, that would require both sufficient physical bandwidth along the way and QoS settings on each link to guarantee the 5 Mbps. You're again correct that normally the tunnel traffic would likely just be treated as BE, but you can treat it special.
E.g. (NB: syntax might be incorrect)
access-list extended 5mbrgre
permit ip host tunnelsrc host tunneldest
class-map match-all 5mbgre
match protocol gre
match group 5mbgre
interface sometype #
service-policy output inbetweenlink
The above example, or something providing the same 5 Mbps guarantee would be needed on all the interfaces, i.e. A to B, B to C, C to D, D to E, and the reverse.
You could also use IP Precedence or DSCP tags on the GRE traffic, and match against those. That would avoid the need for every hop to analyze traffic protocol and/or addresses.