Ideas for a Network Simulation toy..

I've been learning the basics of networking for a few days and suddenly had this idea: to make the model myself in C++.
Though I've not written any code(for this) yet, I think the implementations of the following will be within my capacity(now)...[I'd expand it later along the road]..
* Virtual Machines as a toned down version of the architecture of real comps(in a high level, not 0s and 1s !!)
* Ability to create and destroy new virtual machines(each one being an object).
* Sending and receiving meaningful messages.
* Primitive lookalikes of some network information tools found in Linux and a sort of ssh.
* Illustrating the different topologies and objects resembling the other devices to hold the vm `s together.

Would anyone of you suggest some other ideas &| features, or something which would be a helpful read ?

I don't know how detailed you want your simulation to be, but I'd read OSI model - Wikipedia, the free encyclopedia. Are you going to implement tcp? ipv4/6?

Quote:

---

how detailed you want your simulation to be

---

I was thinking about fusing the lower three layers (upto network) into a single entity(for now)... and the others into another.

Quote:

---

are you going to implement tcp?

---

I'm not sure. I would try to, but if it gets too complicated, I'd make a simple one myself...

Another Question::
Would it be a good idea to learn about threads before attempting to do this ?
__ I mean .. Can I make it in a simple way now, and later break ..say..the different devices of the network, into different threads ...without breaking everything else ? __

Threads would certainly make it more logical. Afterall, in computer networks, computers don't generally run one at a time.

>> Can I make it in a simple way now, and later break ..say..the different devices of the network, into different threads ...without breaking everything else ?

You will always want to have two models; a synchronous and a asynchronous one, to cater for what would be your virtual machine real-life needs. So it's perfectly fine to develop the synchronous model now and later add the asynchronous one.

>> Would anyone of you suggest some other ideas &| features, or something which would be a helpful read ?

I think you covered it. The only other thing I can think off is emulating a minimum set of protocols that will add to your final protocol stack. Each protocol implemented through an object that can only reside on its respective layer. Layers which could exist in the form of a namespace. Dunno... just thinking aloud...

Thank you all..
I've started making the vm part first....
Posting the early design in another thread in the Tech Board ...please do criticize :D