User login

Christopher Lorier's blog

07

Sep

2012

The controller now works with openflow 1.2 with the CPqD of12softswitch.

There is still some stuff that is giving errors with the change, but it is forwarding packets correctly.

31

Aug

2012

With the CPqD switch set up and working, I started debugging my new vlan enabled version of the switch. I found that there was a significantly greater difference between openflow 1.0 and 1.2 than I had first realised.

So after establishing that the problems definitely were with my controller rather than with OPqD or ryu itself, I found that somehow in the process of testing that, I had broken my controller completely.

So I spent the rest of my time pouring of the ryu source trying to work out where the error messages I was getting were coming from.

24

Aug

2012

So today I tried to test the 1.2 version of my switch, but instead got stuck with peripheral tasks. Installing a switch capable of using of1.2 was the biggest of them, but Brendon took care of the issues I was having there..

Just gotta get them to talk to each other now.. The issue seems to be making ryu realise that I want it to talk of1.2 rather than 1.0..

But it is hard to spend fridays not doing what it was you set out to do in the first place.. Might try doing all my wand stuff earlier in the week..

17

Aug

2012

Ok, so I have started rewriting my switch controller for openflow 1.2 to allow it to support vlans.

Ryu doesnt seem to support 1.1, but 1.2 seems to be similar enough that I can use that, and I assume that is why it isnt supported. Though I am not sure, because the spec for 1.2 isnt released yet..

Unfortunately openvswitch doesnt support 1.1 yet, so I will need to use some other software switch, so I am gonna have to look for something.

10

Aug

2012

Ok, so I started implementing Vlans

Openvswitch does not support QinQ so that will have to wait. I have started out implementing it without trunks, and have written it except that when it floods at the moment it floods out all ports, regardless of vlan.

I think to solve this in openflow 1.0 is going to be pretty damn tricky. I may have to look at what is available in openvswitch rather than openflow itself..

Should be fun.

03

Aug

2012

Decided there wasnt much point creating a hacked version of link aggregation, so moved onto vlans instead. Did a bit of reading and planning about that.

27

Jul

2012

This week I fixed up the macs changing ports bug, then added MAC security in a trivial way. So that it immediately drops packets not from the source associated with a port.

Spent a bit of time looking at link aggregation. It seems pretty daunting though.. But I am starting to think it wouldnt take much to get a simple not necessarily ieee standard meeting implementation going.

I think the hardest part of doing it the ieee way is reading the standard...

20

Jul

2012

I fixed the bug in the switch with the help of Joe, and now I have a working implementation of the switch on openvswitch.

Now I am looking at what features I can add to it, and how to do some of those. Considering starting with link aggregation or port security for something nice and easy. Though I am really unsure how the interface for setting them up should work..

There are a couple of bugs to do with macs changing port that I need to straighten out. Dealing with the vms for tests is a bit of a nuisance.

13

Jul

2012

Ok, this week I started out trying to get controllers and openvswitch to speak the same language.

Early on in the week Joe linked me to Ryu, another python openflow controller, which supports nicira. So I had a go with using that instead of pox, and eventually put together a switch that in a kinder, gentler world would would perfectly. So currently trying to understand what cruel ironic twists reality has devised to thwart me.

06

Jul

2012

Ok, in the process of getting openvswitch up in running this week, I managed to break one of the emulation machines, though I am told Joe managed to one-up me on that account..

But, I have everything running. It seems like the resubmit command that exists in pox is not the same as the one that exists in openvswitch, which meant my hacked programme didnt even work, which was a bit of a disappointment. However, I have made a switch which uses wildcards and managed to get it running, though it will produce n^2 rules..

The next step is to write something to add the actions that exist in openvswitch but not in pox to my controller.