@ejs
As I said, it is a thought experiment that hasn't reached a conclusion yet. But I can't find an answer within the expected boundaries, so I'm widening my thinking beyond what I expect to happen. Just to see if it provides any other insights... I understand why you asked about the MTU, but the mere act of you asking is what triggered this train of thought.
I experimented with changing my MTU in the router, and it didn't change the IP Profile reported by the BTW speedtester. Strike 1, but I'll try again later when the line isn't being used for work. DLM fears make experimentation slow here (take note, Mr Grimsley!)
However, this morning I was reading SIN 498, which had this to say (amongst lots of other stuff):
As a result of these overheads, the actual achievable throughput in bits per second is
dependent on the reported VDSL2 rate and frame size of the data being transmitted.
with an example:
For example, if the reported downstream VDSL2 data rate is 40,000 kbit/s and the IP
packet size is 1500 bytes (i.e. Ethernet frame size at End User LAN is 1514 bytes) the
maximum throughput achievable is actually 39,178 kbit/s (when measured at the EU
LAN i.e. no VLAN header, but including Ethernet header). Alternatively, if the IP
packet size is 64 bytes the maximum throughput achievable is further reduced to 35,721
kbit/s. This overhead is particularly important to consider in respect to the downstream
shaper setting on the CP’s BRAS.
What they call "IP Packet Size" sounds awfully similar to the MTU to me.
There's a lot in there for me to read related to tagging, which may affect throughput rates too - and they draw a reference to the framing used for PTM mode, which I haven't yet found a good reference for.
Then I was looking in the BTW FTTC Handbook (issue 12 is the most recent I've got; I'd love a newer one), which broke down throughput vs frame size even further, and has given me some ideas for further MTU experiments. I've attached a snip of the table.
Even those throughput figures aren't the final "IP-level" ones we need, as they probably include the encapsulation headers for PPP and PPPoE, at the very least. The text alludes to some form of traffic management - which might be as simple as some tags being added for QoS - but I need to figure that out too.
I think I need to read more of the two documents ... and not skip so much of the bits I don't understand
Right now, I think that the packet size is important. You used MTU to figure out the TCP and IP overhead that applies above the application 'goodput', but I think it works in the other direction with some of the tagging being added by Openreach/BTW/ISP too.
