Yesterday I joined the SFC meeting, but it seems I had some problems with
Google hangouts since you could not hear me.
Giving the integration of SFC with Genius will be finally done in Carbon, we
are not in a rush, but we should analyze the following topics in the coming
weeks (I think you already agreed on attending to Netvirt meetings every
- ITM tunnel handling when using NSH and when not using NSH. When not
using NSH we might reuse existing ITM tunnels created by Netvirt and when using
NSH we will have to see whether we need a separate set of tunnels for
classification + SFC and regular L2/L3 management handled today by Netvirt.
- Check Genius priorities. Both SFC and Netvirt classifier should work
in the same way and my understanding is that in the work we are doing when it
comes to the integration of SFC with Genius, SFC classification has the same
priority as SFC. I don’t know if that would work with Netvirt’s pipeline, so we
should check if we need a new service registered in Genius or we can reuse any
other existing service.
- Check whether classification rules could not be programmed in all
OVSs. As far as I see it, that would not scale (in a virtualized OPNFV
environment with a couple of compute nodes, tacker is taking 20 seconds to
program very simple classification rules). In an SGi LAN configuration where
classification rules will be programmed dynamically that would not scale.
Another planned feature for Carbon is “VPN service integration”, I think this
topic (how to program classification rules) can be revisited there since
classification must be performed just after receiving the VPN packets.
I know some of these points were touched yesterday, and Brady has already
started some discussions with Daya for some of them, but I’m sure we will have
to dig a bit deeper.
[mailto:sfc-dev-boun...@lists.opendaylight.org] On Behalf Of Brady Johnson
Sent: miércoles, 30 de noviembre de 2016 18:47
To: sfc-dev opendaylight <firstname.lastname@example.org>
Subject: [sfc-dev] Meeting minutes, November 30, 2016
sfc-dev mailing list