User login

Search Projects

Project Members

Joe Stringer admin

OpenFlow 1.1 Support in Open vSwitch

Open vSwitch is a production-quality OpenFlow-controlled software switch. It implements software forwarding of packets based on rules configured by the OpenFlow protocol. The current version of Open vSwitch supports OpenFlow 1.0, which cannot configure behaviour with important protocols such as MPLS and IPv6. More recent versions of the OpenFlow specification support these protocols, but Open vSwitch has yet to implement support for later versions of the specification. The goal of this project is to develop functionality necessary to reasonably claim OpenFlow 1.1 support in Open vSwitch.

30

Apr

2012

Spent a lot of the week working on different assignments, primarily libtrace stuff, but also managed to fix my arbitrary ethernet patch for a few of the broken test cases. Still have 26 test cases to fix up before I can put this patch in for review.

In more interesting news, Open Networking Summit was held a couple of weeks ago over in Santa Clara, CA. The most prominent news was Google coming out in the open about OpenFlow in their datacenters. Indiana University and CPqD also jointly made a demo of a virtualised IP routing stack consisting of NOX, RouteFlow, Quagga and Linux Containers. I worked on RouteFlow over the summer as part of my internship with Google, so it's quite exciting to see it getting some exposure at ONS. (https://sites.google.com/site/routeflow/updates/demoinons2012releasesnew...)

16

Apr

2012

I've been over in Sydney this week for job interviews, which is most of my time spent. Got a bit closer to my initial arbitrary ethernet address masks patch, so next week I hope to start the review process. There's still a few pieces I need to enquire about, in particular some NXM code, and need to get the patch running against the test suite.

09

Apr

2012

Worked on implementing arbitrary Ethernet source & destination address masks this week. Still a little bit slow going with understanding where bits of the code are getting called from, but making some progress. Starting to lag behind what I was hoping to achieve by this point, particularly with daylight savings messing with upstream contact, but I hope to get stuck in during the recess and work towards sending in my first patch.

26

Mar

2012

Spent most of my time this week researching, reading OpenFlow/OpenvSwitch papers and related work. Our project proposals were due this week, so writing up my plans for the project was the main focus. Open vSwitch is a reasonably active project, so there's quite a bit which is up in the air regarding the specifics of how far I'll get with implementing OpenFlow 1.1/1.2 support. For the time being I have a few concrete milestones to work towards, and I'll revisit what I expect to get done in a few weeks' time.

I've made my Honours Proposal available via the following link, for anyone interested in reading more.

http://goo.gl/JXhNW

19

Mar

2012

Met up with Richard to clarify how we're going to organise my project -- Primarily with Richard monitoring progress and making sure I satisfy COMP520, then for technical oversight, making use of the ovs-dev mailinglist, and in particular talking to Ben Pfaff from Nicira. He'll be doing my code reviews and such.

Got in touch with Ben on IRC to get a better overview of the OVS architecture and which parts of the code I'll be working in. Started work on abstracting out the current OFP_MATCH code.

12

Mar

2012

This week was spent mainly just settling back into uni -- getting set up in the lab, going to WAND orientation, taking an initial look at the source code I'll be working with. That sort of thing.

On a positive note, the OpenFlow abstraction patches have now hit master for Open vSwitch -- this will make my work much easier, rather than attempting to apply a dozen patches from the mailinglist onto a local copy of the source, then working from there.