Search Mailing List Archives
[mininet-discuss] Two Simple Questions
j_fra at fastmail.us
Fri Dec 16 09:04:19 PST 2016
I do have some follow-up questions:
(1) Can a user-space protocol stack be run under the ownership of each
mininet host? We need the flexibility to rapidly change protocol
stack internals particular to each host without requiring kernel-
(2) An addition thought is: Can two mininet hosts run separate user
level protocol stacks simultaneously to test interoperability? Say
I write a user-level module that talks over raw sockets and launch
it with a shell command run from the mininet CLI so that it runs on
all hosts, will this allow me to bypass the system network stack?
(3) Are there any sections of the documentation that describe this type
of modification to the mininet system? In other words, how exactly
do I go about avoiding the use of an SDN controller and making sure
that hosts aren't sharing any code above the MAC layer?
If mininet doesn't support this type of change, I'd be happy to help
with contributions. I'm a fairly experienced Python developer.
j_fra at fastmail.us
On Thu, Dec 15, 2016, at 01:59 PM, Md. Islam wrote:
> Yes, you don't have to use SDN. You can set it up with no controller.
> Mininet uses the underlying protocol of your host. So, you can
> experiment custom protocols.
> On Dec 15, 2016 4:07 PM, "Jason Franklin" <j_fra at fastmail.us> wrote:
>> I have two simple questions regarding the usage of Mininet:
>> (1) Do I have to use SDN? That is, can I define a basic Ethernet
>> network with no controller? If so, where can I find this in the
>> documentation? There's nothing in the FAQ about this...
>> (2) Can I run my own custom protocol stack on Mininet hosts? I'd
>> to build a stack starting at raw sockets (conveniently accessible in
>> Python, I believe) and work up from there, injecting faults into the
>> network to test the correctness of the protocols. So is there
>> in Mininet that prevents me from doing this?
>> Just trying to make sure that Mininet is right for my needs. Either
>> way, I must say that it's a great project. Admirable work,
>> for sure!
>> Jason Franklin j_fra at fastmail.us
>> mininet-discuss mailing list mininet-discuss at lists.stanford.edu
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mininet-discuss