[openstack-dev] [Neutron][L3] - L3 High availability blueprint v/s pure L3 packet

A, Keshava keshava.a at hp.com
Thu Jun 5 15:59:21 UTC 2014


In the L3 High availability Blue Print I want to make following observation...

1.       Just before Active going down if there was fragmented packet getting reassembling how to recover it  ?

If that packet is related App, App will try to resend those packet so that it will get reassembled in new board .
                Issue with  packets which are pure L3 level  packets.
               Example  IPSEC packet (where there is recursive  encapsulation) , Neighbor Discover packet, ICMP related which are pure L3 packet.
                                If these packets are get losses during process of reassembled how to recover it ?
                                Worst thing is no one is informing that these packets are lost, because there are no session based Socket for pure L3 level packet.

2.       Since this HA mechanism is stateless (VRRP mechanism) what is the impact those L3 protocol which have status ?

How exactly NAT sessions are handled ?

3.       I think we need to have serious discussion with Distributed functionality (like DVR) for session less High availability solution.

Thanks & Regards,

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140605/7fb65dbb/attachment.html>

More information about the OpenStack-dev mailing list