On: The Ethernet PAUSE frame

[bogomipz]

"The very existence of Ethernet flow control may come as a shock, especially since protocols like TCP have explicit flow control"

Not at all. Ethernet is ancient and there are other transport protocols besides TCP that can and have used it in the past Apple Talk, IPX/SPX, DecNet to name a few. This is the beauty of the OSI model. Ethernet at layer 2 is independent of what rides it at layer 4.

[lmm → bogomipz]

Switches aren't supposed to exist in the OSI model at all, which makes it of questionable usefulness today - when was the last time you used a purely-hubbed network?

[bogomipz → lmm]

I am not understanding the point you are trying to make.

OSI is just a model and it is not limited to end stations. What gives you that impression?

Conceptually and virtually switches most certainly do exist in the OSI model.

For all practical purposes a switch can be thought of as a multiport bridge. And as such it exists at Layer 1 and Layer 2.

[lmm → bogomipz]

> OSI is just a model and it is not limited to end stations. What gives you that impression?

I didn't say anything about end stations? The point is that a switch is inherently a violation of OSI layering (it uses layer-3 information to make layer-2 decisions), which given that practically all modern networks are switched, suggests that the 7-layer model may not be that useful for modelling real-world networks.

[matthiasl → bogomipz]

Pause frames came with 802.3x, in 1997. That's much later than any of the protocols you name, by a decade or two.

[bogomipz → matthiasl]

You missed my point - that ethernet and layer 2 is agnostic to what network or transport layers run on top of it. Those were examples were to illustrate non-TCP transport layers. The date doesn't matter at all.