[openstack-dev] [OVN] [networking-ovn] [networking-sfc] SFC and OVN

Ryan Moats rmoats at us.ibm.com
Tue May 31 17:17:57 UTC 2016



John McDowall <jmcdowall at paloaltonetworks.com> wrote on 05/26/2016 11:08:43
AM:

> From: John McDowall <jmcdowall at paloaltonetworks.com>
> To: Ryan Moats/Omaha/IBM at IBMUS
> Cc: Ben Pfaff <blp at ovn.org>, "discuss at openvswitch.org"
> <discuss at openvswitch.org>, Justin Pettit <jpettit at ovn.org>,
> "OpenStack Development Mailing List" <openstack-
> dev at lists.openstack.org>, Russell Bryant <russell at ovn.org>
> Date: 05/26/2016 11:09 AM
> Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN
>
> Ryan,
>
> My (incomplete) throughts about the flow-classifier are:
>
> 1)  ACL’s are more about denying access, while the flow classifier
> is more about steering selected traffic to a path, so we would need
> to deny-all except allowed flows.
> 2)  The networking-sfc team has done a nice job with the drivers so
> ovn has its own flow-classifier driver which allows us to align the
> flow-classifier with the matches supported in ovs/ovn, which could
> be an advantage.

The ACL table has a very simple flow-classifier structure and I'd
like to see if that can be re-used for the purpose of the SFC classifier
(read that I feel the Logical_Flow_Classifier table is too complex).
My initial thoughts were to look at extending the action column and
using the external-ids field to differentiate between legacy ACLs and
those that are used to intercept traffic and route it to an SFC.

>
> What were your thoughts on the schema it adds a lot of tables and a
> lot of commands – cannot think of anyway around it

In this case, I think that the other tables are reasonable and I'm
uncomfortable trying to stretch the existing tables to cover that
information...

Ryan

>
> Regards
>
> John
>
> From: Ryan Moats <rmoats at us.ibm.com>
> Date: Wednesday, May 25, 2016 at 9:12 PM
> To: John McDowall <jmcdowall at paloaltonetworks.com>
> Cc: Ben Pfaff <blp at ovn.org>, "discuss at openvswitch.org" <
> discuss at openvswitch.org>, Justin Pettit <jpettit at ovn.org>, OpenStack
> Development Mailing List <openstack-dev at lists.openstack.org>, Russell
Bryant <
> russell at ovn.org>
> Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN
>
> John McDowall <jmcdowall at paloaltonetworks.com> wrote on 05/25/2016
> 07:27:46 PM:
>
> > From: John McDowall <jmcdowall at paloaltonetworks.com>
> > To: Ryan Moats/Omaha/IBM at IBMUS
> > Cc: "discuss at openvswitch.org" <discuss at openvswitch.org>, "OpenStack
> > Development Mailing List" <openstack-dev at lists.openstack.org>, Ben
> > Pfaff <blp at ovn.org>, Justin Pettit <jpettit at ovn.org>, Russell Bryant
> > <russell at ovn.org>
> > Date: 05/25/2016 07:28 PM
> > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN
> >
> > Ryan,
> >
> > Ok – I will let the experts weigh in on load balancing.
> >
> > In the meantime I have attached a couple of files to show where I am
> > going. The first is sfc_dict.py and is a representation of the dict
> > I am passing from SFC to OVN. This will then translate to the
> > attached ovn-nb schema file.
> >
> > One of my concerns is that SFC almost doubles the size of the ovn-nb
> > schema but I could not think of any other way of doing it.
> >
> > Thoughts?
> >
> > John
>
> The dictionary looks fine for a starting point, and the more I look
> at the classifier, the more I wonder if we can't do something with
> the current ACL table to avoid duplication in the NB database
> definition...
>
> Ryan
>
> > From: Ryan Moats <rmoats at us.ibm.com>
> > Date: Wednesday, May 25, 2016 at 7:27 AM
> > To: John McDowall <jmcdowall at paloaltonetworks.com>
> > Cc: "discuss at openvswitch.org" <discuss at openvswitch.org>, OpenStack
> > Development Mailing List <openstack-dev at lists.openstack.org>, Ben Pfaff
<
> > blp at ovn.org>, Justin Pettit <jpettit at ovn.org>, Russell Bryant <
> russell at ovn.org
> > >
> > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN
> >
> > John McDowall <jmcdowall at paloaltonetworks.com> wrote on 05/24/2016
> > 06:33:05 PM:
> >
> > > From: John McDowall <jmcdowall at paloaltonetworks.com>
> > > To: Ryan Moats/Omaha/IBM at IBMUS
> > > Cc: "discuss at openvswitch.org" <discuss at openvswitch.org>, "OpenStack
> > > Development Mailing List" <openstack-dev at lists.openstack.org>
> > > Date: 05/24/2016 06:33 PM
> > > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN
> > >
> > > Ryan,
> > >
> > > Thanks for getting back to me and pointing me in a more OVS like
> > > direction. What you say makes sense, let me hack something together.
> > > I have been a little distracted getting some use cases together. The
> > > other area is how to better map the flow-classifier I have been
> > > thinking about it a little, but I will leave it till after we get
> > > the chains done.
> > >
> > > Your load-balancing comment was very interesting – I saw some
> > > patches for load-balancing a few months ago but nothing since. It
> > > would be great if we could align with load-balancing as that would
> > > make a really powerful solution.
> > >
> > > Regards
> > >
> > > John
> >
> > John-
> >
> > For the load balancing, I believe that you'll want to look at
> > openvswitch's select group, as that should let you set up multiple
> > buckets for each egress port in the port pairs that make up a port
> > group.
> >
> > As I understand it, Table 0 identifies the logical port and logical
> > flow. I'm worried that this means we'll end up with separate bucket
> > rules for each ingress port of the port pairs that make up a port
> > group, leading to a cardinality product in the number of rules.
> > I'm trying to think of a way where Table 0 could identify the packet
> > as being part of a particular port group, and then I'd only need one
> > set of bucket rules to figure out the egress side.  However, the
> > amount of free metadata space is limited and so before we go down
> > this path, I'm going to pull Justin, Ben and Russell in to see if
> > they buy into this idea or if they can think of an alternative.
> >
> > Ryan
> >
> > >
> > > From: Ryan Moats <rmoats at us.ibm.com>
> > > Date: Monday, May 23, 2016 at 9:06 PM
> > > To: John McDowall <jmcdowall at paloaltonetworks.com>
> > > Cc: "discuss at openvswitch.org" <discuss at openvswitch.org>, OpenStack
> > > Development Mailing List <openstack-dev at lists.openstack.org>
> > > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN
> > >
> > > John McDowall <jmcdowall at paloaltonetworks.com> wrote on 05/18/2016
> > > 03:55:14 PM:
> > >
> > > > From: John McDowall <jmcdowall at paloaltonetworks.com>
> > > > To: Ryan Moats/Omaha/IBM at IBMUS
> > > > Cc: "discuss at openvswitch.org" <discuss at openvswitch.org>, "OpenStack

> > > > Development Mailing List" <openstack-dev at lists.openstack.org>
> > > > Date: 05/18/2016 03:55 PM
> > > > Subject: Re: [OVN] [networking-ovn] [networking-sfc] SFC and OVN
> > > >
> > > > Ryan,
> > > >
> > > > OK all three repos and now aligned with their masters. I have done
> > > > some simple level system tests and I can steer traffic to a single
> > > > VNF.  Note: some additional changes to networking-sfc to catch-up
> > > > with their changes.
> > > >
> > > > https://github.com/doonhammer/networking-sfc
> > > > https://github.com/doonhammer/networking-ovn
> > > > https://github.com/doonhammer/ovs
> > > >
> > > > The next tasks I see are:
> > > >
> > > > 1. Decouple networking-sfc and networking-ovn. I am thinking that I

> > > > will pass a nested port-chain dictionary holding port-pairs/port-
> > > > pair-groups/flow-classifiers from networking-sfc to networking-ovn.
> > > > 2. Align the interface between networking-ovn and ovs/ovn to match
> > > > the nested dictionary in 1.
> > > > 3. Modify the ovn-nb schema and ovn-northd.c to march the port-
> > chain model.
> > > > 4. Add ability to support chain of port-pairs
> > > > 5. Think about flow-classifiers and how best to map them, today I
> > > > just map the logical-port and ignore everything else.
> > > >
> > > > Any other suggestions/feedback?
> > > >
> > > > Regards
> > > >
> > > > John
> > >
> > > John-
> > >
> > > (Sorry for sending this twice, but I forgot that text/html is not
liked
> > > by the mailing lists ...)
> > >
> > > My apologies for not answering this sooner - I was giving a two day
> > > training on Tues/Wed last week and came back to my son graduating
> > > from HS the next day, so things have been a bit of a whirlwind here.
> > >
> > > Looking at the github repos, I like the idea of passing a dictionary
> > > from networking-sfc to networking-ovn. The flow classifiers should
> > > be relatively straightforward to map to ovs match rules (famous last
> > > words)...
> > >
> > > I've probably missed an orbit here, but in the ovn-northd
implementation,
> > > I was expecting to find service chains in the egress and router
pipelines
> > > in addition to the ingress pipeline (see below for why I think a
service
> > > chain stage in the egress pipeline makes sense ...)
> > >
> > > Also, in the ovn-northd implementation, I'm a little disturbed to see
the
> > > ingress side of the service chain sending packets to output ports - I
> > > think that a more scalable (and more "ovs-like" approach) would be to
> > > match the egress side of a port pair in the chaining stage of the
> > > ingress pipeline, with an action that  set the input port register.
> > > Then the egress pipeline would have a chaining stage where the output
> > > port register would be set based on the ingress port of the next port
> > > pair in the chain and the packet being punted to the proper output
port
> > > in the last table.  That should automagically build your function
chain
> > > and provide the basis for bucketizing multiple ingress ports for the
> > > next port group to support hash based load balancing.
> > >
> > > Does that make sense?
> > >
> > > Ryan[attachment "ovn-nb.ovsschema.sfc" deleted by Ryan Moats/
> > Omaha/IBM] [attachment "sfc_dict.py" deleted by Ryan Moats/Omaha/IBM]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160531/a1fd5fc5/attachment-0001.html>


More information about the OpenStack-dev mailing list