Re: [Roll] Clarification on the objective function

["Julien Abeille (jabeille)" <jabeille at cisco.com>]

Hi Pascal,

 

just to be sure I understand: 

- I have a preferred parent at rank 1, my rank is 5, I receive a DIO for a parent rank 2, must i ignore it? I would say no as this node is useful as backup parent, but if i lose my preferred parent, choosing the backup as preferred will make my rank increase (this is ok because if i compute rank based on all parents, I would have to increase rank at reception of the second DIO anyway)

- At boot time, I receive two DIOs (rank 1 and 2) before computing OCP. If I calculate rank based on preferred only, if at some point I would like to switch to the other parent, I will have to move down.

 

Still I think it is preferable to calculate rank based on preferred only, but what do other think?

Julien

---

From: roll-bounces at ietf.org [mailto:roll-bounces at ietf.org] On Behalf Of Pascal Thubert (pthubert)
Sent: lundi 9 novembre 2009 13:39
To: JP Vasseur (jvasseur); Mathilde Durvy (mdurvy)
Cc: roll
Subject: Re: [Roll] Clarification on the objective function

Hi Mathilde,

 

If that parent is preferred, it is probably because that?s where most of the traffic should land.

This is why rank and metrics are expected to be propagated from that of that preferred parent.

 

Allowing other parents to impact the rank is opening the door to greediness. So we picked the simple path of accepting parents that are upwards once the rank is selected from the honest-to-God preferred parent.

 

Pascal

From: roll-bounces at ietf.org [mailto:roll-bounces at ietf.org] On Behalf Of JP Vasseur (jvasseur)
Sent: mardi 3 novembre 2009 12:30
To: Mathilde Durvy (mdurvy)
Cc: roll
Subject: Re: [Roll] Clarification on the objective function

 

Hi,

 

On Nov 3, 2009, at 11:28 AM, Mathilde Durvy (mdurvy) wrote:

Hi All,

 

Section 5.9.1 says

     "When the scan is complete, the preferred parent is elected and
      self's rank is computed as the preferred parent rank plus the step
      in rank with that parent."
Does this mean that the rank of a node is only based on the rank of its most preferred parent?

Shouldn't the rank of the node also depend on the non-preferred parents?

Or is it assumed that the node always forward traffic via its most preferred parent? 

 

A node may indeed choose to send traffic to an alternate parent and not the most preferred.

But do you think that this should change the way its computes its rank ?

 

JP.

 

Best,

Mathilde

 

Durvy Mathilde Software Engineer Technology center mdurvy at cisco.com Phone: +41 21 822 1725 Mobile: +41 76 396 8116	Cisco Systems International Sàrl Av. des Uttins, 5 CH-1180 Rolle Cisco home page

 

Think before you print.

This e-mail may contain confidential and privileged material for the sole use of the intended recipient. Any review, use, distribution or disclosure by others is strictly prohibited. If you are not the intended recipient (or authorized to receive for the recipient), please contact the sender by reply e-mail and delete all copies of this message.

 

_______________________________________________
Roll mailing list
Roll at ietf.org
https://www.ietf.org/mailman/listinfo/roll