Hi Sasha,
Sorry for the late relpy.
The two options currently provided in the draft are not applicable for any
topology, but only for clos/fat tree. And the sending of ARN messages is
directional, ARN will only be sent to the upstream device that has forwarded
the message upstream.
So for the remote notification case, when there's no other non-congested path
available on the directly connected upstream device, ARN is sent to the
non-directly connected device, and after this device reroutes the traffic to a
new next hop, there may be three situations:
1）The next-hop device forwards the traffic to the destination through an
uncongested path. -- Goal achieved
2）The next-hop device does not have an available forwarding entry for the
traffic. --- Additional mechanisms are needed to avoid black hole/packet
dropping, e.g, via the controller.
3）The next-hop device forwards the traffic to the congested device again. ----
The congested device will generate ARN for the traffic again and send it to the
upstream, aiming to trigger the redirection of the traffic again. The result
would be 1) 2) or 3), but since only the upstream nodes might reroutes the
traffic, it would not cause a loop.
Loop may occur in situation 1) and 3) when the next-hop device
changes/recomputes it's routing table and sends the traffic back to the
upstream node due to link changes in the network. But this is a loop during
routing convergence, which also occurs in normal scenarios, and a micro-loop
avoidance mechanism is needed.
Thanks,
Yao
Original
From: AlexanderVainshtein <alexander.vainsht...@rbbn.com>
To: draft-liu-rtgwg-adaptive-routing-notificat...@ietf.org
<draft-liu-rtgwg-adaptive-routing-notificat...@ietf.org>;
Cc: rtgwg@ietf.org <rtgwg@ietf.org>;
_______________________________________________
rtgwg mailing list -- rtgwg@ietf.org
To unsubscribe send an email to rtgwg-le...@ietf.org
Hi all,
Just to clarify: my question at the mike was about perceived lack of any loop
avoidance mechanisms in the Adaptive Routing Notification for Load-balancing
draft.
I have to admit that I did not fully understand the response of the author and
suggest discussing this issue Furter on the list.
Regards,
Sasha
Disclaimer
This e-mail together with any attachments may contain information of Ribbon
Communications Inc. and its Affiliates that is confidential and/or proprietary
for the sole use of the intended recipient. Any review, disclosure, reliance or
distribution by others or forwarding without express permission is strictly
prohibited. If you are not the intended recipient, please notify the sender
immediately and then delete all copies, including any attachments.
_______________________________________________
rtgwg mailing list -- rtgwg@ietf.org
To unsubscribe send an email to rtgwg-le...@ietf.org
