R: [Netkit.users] Netkit benchmarking

[Massimo Rimondini]

Dear Paolo,

I'm glad you hear the explanation was useful!

I think the answer to all your questions is unique.
A Netkit lab consists of several processes (virtual machines, virtual hubs,
etc.), and the sharing of CPU power among them is regulated by the host
kernel scheduler. Hence, the sharing policy depends on the host kernel you
are running and the parameters it has been compiled with.
We don't have benchmarks because Netkit is an emulation environment, in
which accuracy matters more than speed. Even if we provided them, they would
be highly biased by both the hardware and the software (i.e., kernel) run on
the host machine, and therefore we believe they wouldn't be of much use.

About the packet loss, the message you see is signalled by the software that
manages virtual connections. In short, it is a small program which
dispatches the packets sent by virtual machines on a Unix socket. If packets
are lost, this is usually because one of the virtual machines is unable to
eat them up at a proper rate. The fact that this happens while performing a
scan is likely to be related to this.

...and don't worry about your english :-)

Regards,
Massimo.


> -----Messaggio originale-----
> Da: netkit.users-bounces a list.dia.uniroma3.it 
> [mailto:netkit.users-bounces a list.dia.uniroma3.it] Per conto 
> di Paolo Carpo
> Inviato: lunedì 11 settembre 2006 16.56
> A: netkit.users a list.dia.uniroma3.it
> Oggetto: [Netkit.users] Netkit benchmarking
> 
> Hi all!
> 
> First of all, thanks a lot to Massimo Rimondini for the long 
> explanation on some dubts some time ago: now I can run Netkit 
> with 2 vmware hosts in different subnets on the same pc, cool ;)
> 
> But I have other questions: while there isn't the capability 
> to define the link bandwidth and delay, I would like to know 
> if there are some benchmark to know how much a connection 
> between 2 Netkit nodes can carry. If I try hping or even 
> nmap, the communication losts some packets (Netkit returns: 
> send_sock sending to fd 4 Resource temporarily unavailable).
> 
> Another question, in the same contest, is: do you have made 
> some bench on the CPU of the VM? How the CPU power of the 
> real PC is divided between the VMs?
> 
> I'm afraid my english is really terrible, but I hope it's 
> clear enough.
> 
> Thank you very much in advice,
> 
> 	Paolo Carpo 
>