Re: [Roll] question about broken link detection

[[e-mail]]

Hi Jorjeta,

The P2P mechanism could be deployed as a band-aid means to repair the link in the DAG, but it is not meant for that.  If node A wants to transmit to node B within the LLN, RPL (in non-storing mode) would require the packet to traverse to the LBR and then be source routed back down to the destination.  If the packet should happen upon a broken link, it would be feasible that a P2P request could be made at that point to re-establish a link and complete the journey.  However since P2P is not an integral part of RPL, this would be a proprietary solution.  The packet would get to the destination, but the LBR would not know about the broken link so there would be no persistence of this link repair.

The more likely scenario would be that Node A sends a packet to Node B via the LBR as noted above.  Node A times out the Node B response and determines the packet has been lost.  Rather than continue to retry through the DAG and LBR hoping the link has been re-established; Node A rather defines a direct link to Node B using the P2P mechanism.  Here, rather than traversing the entire DAG up and down through the LBR, the packet can be efficiently sent directly from A to B through the appropriate hopping nodes.

This is the intent of the P2P draft.  Hope this helps.

Jerry

From:	"Don Sturek" <[e-mail]>
To:	"'Jetcheva, Jorjeta'" <[e-mail]>, <[e-mail]>
Date:	01/06/2011 08:20 AM
Subject:	Re: [Roll] question about broken link detection

---

Hi Jorjeta,
 
I think the point to point draft (http://datatracker.ietf.org/doc/draft-dt-roll-p2p-rpl/" rel="nofollow">http://datatracker.ietf.org/doc/draft-dt-roll-p2p-rpl/) attempts to take up the issue of pro-active route repair using RPL.   I have not closely followed the draft but know that was in scope.
 
Don
 
 
From: [e-mail] [mailto:[e-mail]] On Behalf Of Jetcheva, Jorjeta
Sent: Monday, January 03, 2011 11:14 AM
To: [e-mail]
Subject: [Roll] question about broken link detection
 
Hi everyone,
 
Happy New Year!
 
I am trying to understand how RPL deals with broken links in non-storing mode so I wanted to run several scenarios by the group.  When a packet is transmitted downstream and reaches a broken link, e.g., from node A to node B,  and the transmitting node (node A) doesn’t get a link layer ack from the downstream next hop (node B) after sending several retransmissions, it can send an ICMP destination unreachable packet to the root.  The root can then try to use other routes it has to the destination or if it doesn’t know of any other ones, it can:
 
1.        Increment its DTSN, triggering all nodes in the DODAG to send it a DAO, at which point the downstream node adjacent to the broken link (node B) will detect that the link to its parent (node A) is broken and will select a new preferred parent (assuming it has other possible parents) and send a DAO.
2.       Wait for the downstream node in the broken link (node B) to detect that its parent (node A) is not reachable
A.      When the downstream node (Node B) observes that it hasn’t gotten a DIO from its parent (node A) for a while, and it would select a new preferred parent (assuming it has other possible parents) and send a DAO to the root
B.      If traffic happens to be flowing upstream, the lack of per-hop/link layer acks would indicate the upstream parent (node A) is no longer reachable and a new parent will be selected, and a DAO would be sent to the root.
Are these the options for detecting a broken link when forwarding a packet downstream in non-storing mode?  Is there a preferred mechanism for this?
 
Thanks!
 
Jorjeta_______________________________________________
Roll mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/roll" rel="nofollow">https://www.ietf.org/mailman/listinfo/roll