The Myth Of Linux Community Support. 960

Let's try to make this very simple. A protocol is a kind of simple signalling and data transfer system. As you say you are a programmer, you could look at it this way - it's a programme which is operated by two machines, taking turns, much like having a conversation. The *only* thing you can do on a network is move data of one kind or another between two points (okay, you can broadcast, but that's not bi-directional).

What makes protocols useful is what each end does with the data, therefore, the specific contents of the packets or circuits are of *no* consequence at all, they need the context of the processes managing the link at the end-points.

So, in order to move data, you need a process at each end. It's precisely the same whether your protocol is operating at a very high layer, like X, or at a lower layer, like tcp or ip or sdh. There is no fundamental difference between an application server or a file server, they are doing the same thing.

You are attempting to put significance to what the processes at the end do with the data on the basis of what moves between the points. You've claimed that there is a difference between 'mounting' a remote - this is just plain wrong. You can claim that being at a higher or lower layer (eg. the end points are user applications or kernel daemons) makes them completely different, but that, to me, is a strawman argument, a term you seem to use rather too frequently.

Networking really is very simple, if you think it's inherently complex then re-read, because it's not. The devil's in the detail, I'll grant, but the principles are simple.

-- Mark Kent -- mark at ellandroad dot demon dot co dot uk