The MPLS WG Archive

[JP Vasseur <jvasseur@cisco.com>]

Hi,

On Jun 21, 2005, at 5:24 AM, Balaji, Hemanth wrote:

>
>
> Hi,
>
> Please Excuse me. There still seemed to have been some alignment
> problem.
>
> -Hemanth.
>
> -----Original Message-----
> From: mpls-bounces@lists.ietf.org [mailto:mpls-bounces@lists.ietf.org]
> On Behalf Of Balaji, Hemanth
> Sent: Tuesday, June 21, 2005 11:03 AM
> To: Adrian Farrel; George Swallow; mpls@ietf.org
> Subject: RE: [mpls] One week last-last call on Soft Preemption
>
>
>
> Hi,
>
>  Can anyone explain, how the following situation is handled ( be it
> needed to ) by the soft-preemption draft.
>
>
>
>
> Topology:
>
>                    +---------------- R2
>                    |
> R0 --------------- R1
>                    |
>                    +---------------- R4
> Reservable link Bandwidth( BW)  for all links = 1000
>
>
> LSP1 :
>
> R0 ----------------- R1 ---------------- R4
>
> BW = 500
> Priority(Hold): 7
>
>
>
>
> LSP2 :
>
> R2 ----------------- R1 ---------------- R4
>
> BW = 600
> Priority(Setup): 6
>
> 1) Initially LSP1 is UP. Soft preemption desired flag is set.
>
> 2) Now R1 gets a new resquest for LSP2. ( which requires LSP1 to be
> preempted, softpreempted in this case)
>
> 3) Let us assume the following situation.
>
>
>> Before R1 notifies R0 about the preemption.
>> R0 changes its BW Req to 400 and sends out its first new path
>>
>      message.
>
>> The first Resv msg from R1 that R0 receives has (will have-due to
>>
>      step 2) the preemption pending flag set.
>
>
> Now, what will the behavior be in this case, will this situation make
> any difference?
>
> Ideally, R0 should not attempt a make before break and look for an
> allternate path, since the new BW requirement ( 400(7) + 600(6) )
> can be satisfied along the same path ( R0-R1-R4).
>

Looks like your question is not really related to soft preemption but  
the head-end reroute engine to find a new path for a (soft) preempted  
TE LSP. The head-end should avoid this path indeed. That said, even  
if it retries the same path (in your example), the LSP setup will not  
complete ...

JP.

>
> Regards,
> Hemanth.
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
> mpls mailing list
> mpls@lists.ietf.org
> https://www1.ietf.org/mailman/listinfo/mpls
>

_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls