Open Stack

Attendees (Sorry we did not catch all the names):
Cathy Zhang (Huawei), Louis Fourie, Alex Barclay (HP), Vikram Choudhary, Carlos Goncalves (NTT) mail at cgoncalves.pt<mailto:mail at cgoncalves.pt>, Adolfo Duarte (HP) adolfo.duarte at hp.com<mailto:adolfo.duarte at hp.com>
German Eichberger (HP)  german.eichberger at hp.com<mailto:german.eichberger at hp.com>, Swami Vasudevan (HP)  swaminathan.vasudevan at hp.com<mailto:swaminathan.vasudevan at hp.com>, Uri  Elzur (Intel)  uri.elzur at intel.com<mailto:uri.elzur at intel.com>
Joe D'Andrea (ATT) jdandrea at research.att.com<mailto:jdandrea at research.att.com>, Isaku Yamahata (Intel) isaku.yamahata at gmail.com<mailto:isaku.yamahata at gmail.com>, Malini Bhandaru malini.k.bhandaru at intel.com<mailto:malini.k.bhandaru at intel.com>,
Michael Johnson (HP) johnsom at hp.com<mailto:johnsom at hp.com>, Lynn Li (HP) lynn.li at hp.com<mailto:lynn.li at hp.com>, Mickey Spiegel (IBM) emspiege at us.ibm.com<mailto:emspiege at us.ibm.com>, Ryan Tidwell (HP) ryan.tidwell at hp.com<mailto:ryan.tidwell at hp.com>
Ralf Trezeciak openstack at trezeciak.de<mailto:openstack at trezeciak.de>, David Pinheiro, Hardik Italia

Agenda:
Cathy presented the service chain architecture and blueprints and Gerrit review for:

-           Neutron Extensions for Service chaining

-          Common Neutron SFC Driver API



There is a lot of interest on this feature.



Questions

Uri: Does this BP specify the implementation in the backend data path? Cathy: This could be implemented by various data path chaining schemes: eg. IETF service chain header, VLAN.

Uri: What happens if the security group/iptables conflict with SGs for port-chain? Cathy: it is expected that conflicts will be resolved by the upper layer "Intent Infra".

Vikram: can other datapath transports be used. Cathy: Yes, what is defined in the BP is the NBI for Neutron.

Swami: add use case and example. Cathy: will add. There is a SFC use case BP, will contact authors.

Carlos: how does this relate to older traffic steering BP. Cathy: we can discuss offline and work together to incorporate the traffic steering BP idea into this BP.

Uri: are these two BPs for NBI and SBI for Neutron? Cathy: yes

Isaku: does Kyle know about BPs? Cathy: yes

Uri: how do these BPs relate to rest of Openstack? Cathy: There will be a presentation on the service chain framework at 2pm May 18 at the Vancouver Summit. It will be presented by Cathy together with Kyle Mestery and Dave Lenrow.

Swami:  It helps to present a simple service chain use case at meeting next week. Suggest to have IRC meeting too as it provides a record of what is discussed. Cathy: I have already created the IRC meeting for service chain project, will start the IRC meeting too.

Malini: we should have a complete spec so redesign is avoided.

Swami: We should use Neutron design session at Vancouver to discuss this BP and flesh out details.
Action Items


1.     We will have two meetings next week: one "goto" meeting with audio and data sharing, the other IRC meeting with link to google doc for sharing diagrams. Cathy will send out the meeting info to the community

2.     Swami will send an email to Kyle Mestery requesting a time slot for service chaining topic in Neutron design session so that we can get all interested parties in one room for a good face-to-face discussion.

3.     Cathy will discuss with Carlos offline about incorporating the traffic steering idea into the service chain BPs.

4.     Service chaining is a complicated solution involving management plane, control plane, and data plane as well as multiple components. The consensus is to first design/implement the two Neutron related service chain BPs https://review.openstack.org/#/c/177946 and get the feature approved by Neutron Core/Driver team and implemented for the OpenStack L release. If we can not get a slot in the design session, we will meet at the service chain presentation on May 18 and find a room to discuss how to move forward with this feature development in OpenStack.

Feel free to chime in if I miss any points.

Thanks,
Cathy



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150505/cd9113eb/attachment.html>