<html><body><p><tt>John McDowall <jmcdowall@paloaltonetworks.com> wrote on 05/25/2016 07:27:46 PM:<br><br>> From: John McDowall <jmcdowall@paloaltonetworks.com></tt><br><tt>> To: Ryan Moats/Omaha/IBM@IBMUS</tt><br><tt>> Cc: "discuss@openvswitch.org" <discuss@openvswitch.org>, "OpenStack <br>> Development Mailing List" <openstack-dev@lists.openstack.org>, Ben <br>> Pfaff <blp@ovn.org>, Justin Pettit <jpettit@ovn.org>, Russell Bryant<br>> <russell@ovn.org></tt><br><tt>> Date: 05/25/2016 07:28 PM</tt><br><tt>> Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN</tt><br><tt>> <br>> Ryan,</tt><br><tt>> <br>> Ok – I will let the experts weigh in on load balancing.</tt><br><tt>> <br>> In the meantime I have attached a couple of files to show where I am<br>> going. The first is sfc_dict.py and is a representation of the dict <br>> I am passing from SFC to OVN. This will then translate to the <br>> attached ovn-nb schema file.</tt><br><tt>> <br>> One of my concerns is that SFC almost doubles the size of the ovn-nb<br>> schema but I could not think of any other way of doing it.</tt><br><tt>> <br>> Thoughts?</tt><br><tt>> <br>> John</tt><br><br><tt>The dictionary looks fine for a starting point, and the more I look</tt><br><tt>at the classifier, the more I wonder if we can't do something with</tt><br><tt>the current ACL table to avoid duplication in the NB database</tt><br><tt>definition...</tt><br><br><tt>Ryan</tt><br><tt><br>> From: Ryan Moats <rmoats@us.ibm.com><br>> Date: Wednesday, May 25, 2016 at 7:27 AM<br>> To: John McDowall <jmcdowall@paloaltonetworks.com><br>> Cc: "discuss@openvswitch.org" <discuss@openvswitch.org>, OpenStack <br>> Development Mailing List <openstack-dev@lists.openstack.org>, Ben Pfaff <<br>> blp@ovn.org>, Justin Pettit <jpettit@ovn.org>, Russell Bryant <russell@ovn.org<br>> ><br>> Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN</tt><br><tt>> <br>> John McDowall <jmcdowall@paloaltonetworks.com> wrote on 05/24/2016 <br>> 06:33:05 PM:<br>> <br>> > From: John McDowall <jmcdowall@paloaltonetworks.com><br>> > To: Ryan Moats/Omaha/IBM@IBMUS<br>> > Cc: "discuss@openvswitch.org" <discuss@openvswitch.org>, "OpenStack <br>> > Development Mailing List" <openstack-dev@lists.openstack.org><br>> > Date: 05/24/2016 06:33 PM<br>> > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN<br>> > <br>> > Ryan,<br>> > <br>> > Thanks for getting back to me and pointing me in a more OVS like <br>> > direction. What you say makes sense, let me hack something together.<br>> > I have been a little distracted getting some use cases together. The<br>> > other area is how to better map the flow-classifier I have been <br>> > thinking about it a little, but I will leave it till after we get <br>> > the chains done.<br>> > <br>> > Your load-balancing comment was very interesting – I saw some <br>> > patches for load-balancing a few months ago but nothing since. It <br>> > would be great if we could align with load-balancing as that would <br>> > make a really powerful solution.<br>> > <br>> > Regards<br>> > <br>> > John<br>> <br>> John-<br>> <br>> For the load balancing, I believe that you'll want to look at<br>> openvswitch's select group, as that should let you set up multiple<br>> buckets for each egress port in the port pairs that make up a port<br>> group.<br>> <br>> As I understand it, Table 0 identifies the logical port and logical<br>> flow. I'm worried that this means we'll end up with separate bucket<br>> rules for each ingress port of the port pairs that make up a port<br>> group, leading to a cardinality product in the number of rules.<br>> I'm trying to think of a way where Table 0 could identify the packet<br>> as being part of a particular port group, and then I'd only need one<br>> set of bucket rules to figure out the egress side. However, the<br>> amount of free metadata space is limited and so before we go down<br>> this path, I'm going to pull Justin, Ben and Russell in to see if<br>> they buy into this idea or if they can think of an alternative.<br>> <br>> Ryan<br>> <br>> > <br>> > From: Ryan Moats <rmoats@us.ibm.com><br>> > Date: Monday, May 23, 2016 at 9:06 PM<br>> > To: John McDowall <jmcdowall@paloaltonetworks.com><br>> > Cc: "discuss@openvswitch.org" <discuss@openvswitch.org>, OpenStack <br>> > Development Mailing List <openstack-dev@lists.openstack.org><br>> > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN<br>> > <br>> > John McDowall <jmcdowall@paloaltonetworks.com> wrote on 05/18/2016 <br>> > 03:55:14 PM:<br>> > <br>> > > From: John McDowall <jmcdowall@paloaltonetworks.com><br>> > > To: Ryan Moats/Omaha/IBM@IBMUS<br>> > > Cc: "discuss@openvswitch.org" <discuss@openvswitch.org>, "OpenStack <br>> > > Development Mailing List" <openstack-dev@lists.openstack.org><br>> > > Date: 05/18/2016 03:55 PM<br>> > > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN<br>> > > <br>> > > Ryan,<br>> > > <br>> > > OK all three repos and now aligned with their masters. I have done <br>> > > some simple level system tests and I can steer traffic to a single <br>> > > VNF. Note: some additional changes to networking-sfc to catch-up <br>> > > with their changes.<br>> > > <br>> > > <a href="https://github.com/doonhammer/networking-sfc">https://github.com/doonhammer/networking-sfc</a> <br>> > > <a href="https://github.com/doonhammer/networking-ovn">https://github.com/doonhammer/networking-ovn</a><br>> > > <a href="https://github.com/doonhammer/ovs">https://github.com/doonhammer/ovs</a> <br>> > > <br>> > > The next tasks I see are:<br>> > > <br>> > > 1. Decouple networking-sfc and networking-ovn. I am thinking that I <br>> > > will pass a nested port-chain dictionary holding port-pairs/port-<br>> > > pair-groups/flow-classifiers from networking-sfc to networking-ovn.<br>> > > 2. Align the interface between networking-ovn and ovs/ovn to match <br>> > > the nested dictionary in 1.<br>> > > 3. Modify the ovn-nb schema and ovn-northd.c to march the port-<br>> chain model.<br>> > > 4. Add ability to support chain of port-pairs<br>> > > 5. Think about flow-classifiers and how best to map them, today I <br>> > > just map the logical-port and ignore everything else.<br>> > > <br>> > > Any other suggestions/feedback?<br>> > > <br>> > > Regards<br>> > > <br>> > > John<br>> > <br>> > John-<br>> > <br>> > (Sorry for sending this twice, but I forgot that text/html is not liked<br>> > by the mailing lists ...)<br>> > <br>> > My apologies for not answering this sooner - I was giving a two day<br>> > training on Tues/Wed last week and came back to my son graduating<br>> > from HS the next day, so things have been a bit of a whirlwind here.<br>> > <br>> > Looking at the github repos, I like the idea of passing a dictionary<br>> > from networking-sfc to networking-ovn. The flow classifiers should<br>> > be relatively straightforward to map to ovs match rules (famous last<br>> > words)...<br>> > <br>> > I've probably missed an orbit here, but in the ovn-northd implementation,<br>> > I was expecting to find service chains in the egress and router pipelines<br>> > in addition to the ingress pipeline (see below for why I think a service<br>> > chain stage in the egress pipeline makes sense ...)<br>> > <br>> > Also, in the ovn-northd implementation, I'm a little disturbed to see the<br>> > ingress side of the service chain sending packets to output ports - I<br>> > think that a more scalable (and more "ovs-like" approach) would be to<br>> > match the egress side of a port pair in the chaining stage of the<br>> > ingress pipeline, with an action that set the input port register.<br>> > Then the egress pipeline would have a chaining stage where the output<br>> > port register would be set based on the ingress port of the next port<br>> > pair in the chain and the packet being punted to the proper output port<br>> > in the last table. That should automagically build your function chain<br>> > and provide the basis for bucketizing multiple ingress ports for the<br>> > next port group to support hash based load balancing.<br>> > <br>> > Does that make sense?<br>> > <br>> > Ryan[attachment "ovn-nb.ovsschema.sfc" deleted by Ryan Moats/<br>> Omaha/IBM] [attachment "sfc_dict.py" deleted by Ryan Moats/Omaha/IBM] </tt><BR>
</body></html>