[Netkit.users] VDE & Netkit

[Antonio Anselmi]

Hello everybody! spring is coming .....and job still remains!

----- Original Message -----
From: "Massimo Rimondini" <rimondin a dia.uniroma3.it>
To: <netkit.users a dia.uniroma3.it>
Sent: Tuesday, April 26, 2005 6:27 PM
Subject: Re: [Netkit.users] VDE & Netkit

> Antonio Anselmi wrote:
>
[snip]
> >Maybe we can think to a particular option, recognised by the vstart,
which
> >permits to commute betwen uml_switch or vde_switch ...

> That's exactly what I have been thinking about. Currently, users can
> configure the physical network topology based on the "metaphor" of
> collision domains: every uml_switch corresponds to a collision domain,
> and interfaces are attached to these "virtual hubs". A possible future
> scenario is the one in which all the switches are vde_switch'es, and
> users can optionally choose to attach an interface to a "remote"
> collision domain (something like: --eth0=192.168.161.3:dom_A), provided
> that they have an account on the remote machine.
> Of course, this would only support basic usage of the VDE architecture,
> but it's just a very preliminary idea. And, you know: the more the
> flexibility, the less the user friendliness ;)

yes indeed Massimo! let me talk about this idea ....
*I'll be a little bit descriptive for people in this list who don't know
vde*

V-clients who want share a collision domain located on a remote host have to
open a secure tunnel and run dpipe inside it, plugging in two vde-plugs at
the sides of
the tunnel:

dpipe vde_plug <channel> <our_socket> = ssh <remote_machine> vde_plug
<channel>  <remote_socket>

Obviously, users must know the couple "socket name-channel" at their
disposal on the other side (other than an account): in other words, the
remote master host must divulge a kind of a "well-know port number" (socket
name-channel) for the inbound vde connections.
Makink the right thing, the remote master host should also notice the
clients about their reserved IP-MAC address to use too! (no DHCP at present
time).
The remote master host may mantain a kind of /etc/virtual-hosts as this:

- vm's physical IP source address
- virtual MAC and IP assigned
- socket and channel assigned
- nick name assigned

The second and the last filelds may be shareable among the users (by eMail
...like the beginning times!).
In this scenario, as well understood, an ordinary account is no more
sufficient: this is to say that an user needs distinct "accounting packet"
for every remote host to connect, consisting in:

- remote host routable address (or its FQDN)
- username and password
- name of remote socket and relative channel number
- IP-MAC address assigned
- nick name (not necessary)

This "accounting packet" mey be autenticated matching the correspondent line
in /etc/virtual-hosts.

> You appear to be considering a very large scenario. I guess the only way
> to ensure MAC addresses are unique would be to request them from a
> (centralized?) service (think about a "virtual provider" that assigns
> IP/MAC addresses - of course, for free!).

I appreciate this idea ...a good idea indeed! Read me: we really can begin
some experiences about what I have wrote above.
What I propose is this:
dia uniroma3 act as  the above remote-master-host, who:

- publish the name of the socket assigned to its vde_switch
- assign channel number, MAC and (non routable) IP address for every virtual
machine wo want share  the switch
- assign an ordinary account for ssh (maybe by traditional mail...)
- establish a period of time in witch the connection are possible (08-10 AM,
for example)

So we realize the "virtual square" as introduced by Renzo Davoli -
University of Bologna - coder and mantainer for vde (Marco Fiscato can
cantact him).
In another way (due to possible security problems at dia uniroma3) everyone
may post to each user (who asked for) its own parameters "Hello guy: this is
may IP for ssh, this is your account, channel and socket name. You must
introduce your vm to my switch with this IP and MAC. Remember, the
connection remains open from 3 to 5 PM ! have fun)".

THINK: we all in this list have the opportunity to realize a nice virtual
"wide collision domain"  where make experiences connecting our own virtual
machines (UML, netkit, qemu,...), we can set-up a virtual segment inside
internet which appears and disappears ...(better than old napster!).

As Lennon  ..."you may say I am a dreamer" , but I think this scenario is
not a futuristic one, I think this is within easy reach of us.

What is the opinion of dia uniroma3 ?
I can offer the above info in order to realize a vde-connection to my
switch, anyone can do the same ?
Let me know your suggestions, comments, ideas  ....

Bye,
Antonio

P.S.

> Anyway, I think equipping *_switches with the capability to emulate this
> results in simpler and more flexible configuration of the virtual network.
>
Ok Massimo, you're in the right side! the "job" was reffered to a pure and
simpy switch!