Search Mailing List Archives
[mininet-discuss] ovs-vswitchd support?
brandonh at stanford.edu
Tue Oct 25 09:46:22 PDT 2011
It would be great if you could try to spin one up. OVS is certainly moving
away ovs-openflowd and it would be nice to have MN install out of the box on
OVS 1.2+, without using deprecated stuff.
The possible downside is that if I understand it, OVS w/vswitchd starts up a
database process, which at least sounds heavier-weight than the
ovs-openflowd. So rather than completely deprecate it immediately, I would
want to see a test that measures practical effect on startup time for
increasing numbers of switches, w/a patch that allows user choice until any
decision is made. This could be a simple shell script with a for loop that
runs something link 'time mn --ovs_vswitchd --topo linear,x' for different
numbers of x.
If you can send the patch over, I'd be happy to test it. If you go ahead,
please fork from the mininet master branch, which is now on GitHub. If you
send a pull request, we can even try out GH's code review system:
On Mon, Oct 24, 2011 at 9:58 PM, Rob Sherwood <rob.sherwood at bigswitch.com>wrote:
> Hi all,
> Poking through the mininet archives, I've seen a few people asking
> about ovs-vswitchd support in mininet: did anyone actually make a
> If no, I think I will try to spin one up, because at least according
> to the NEWS file in OVS, the ovs-openflowd method is no longer being
> supported (and has been deprecated to "test-openflowd" in 1.2 and post
> 1.2 has even been removed).
> Bob, Brandon: what are your thoughts about me completely deprecating
> the ovs-openflowd and making the 'ovsk' mininet switch target just
> invoke ovs-vswitchd (and the corresponding database process,
> ovsdb-server) as the ovs folks now prescribe?
> Please let me know and thanks,
> - Rob
> mininet-discuss mailing list
> mininet-discuss at lists.stanford.edu
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mininet-discuss