@Kitz
"This is confirmed as definite"
Is that confirmation based on seeing the traceroute? Or an independent confirmation via other means?
Initially I suspected because of the cease. Once I'd seen the tracert it was a pretty sure bet.
It was confirmed last night by other means. Once I'd sussed it and said so, there was no point denying it otherwise they would have just looked stupid.
Is the lack of a bRAS/backhaul for nodes A and B significant? Or just a shortcut, and we should assume they exist there too?
Yes, its a short cut to show that they will have other nodes.
The backhaul, bRAS, BTRADIUS etc will still be there as will the MSANs (or DSLAMs in the case of fttc).
stating that things won't always be visible via traceroute. I wonder why not?
No idea sorry that does seem a little strange - as does the fact atm how both the core node by the RAS and the end node shows an ip as part of a private network. Whether in time & once all the nodes are live they may allocate external IPs to show proper ownership I dont know. They have a lot more flexibility to do what they like,
One thought does spring to mind though & jelv will perhaps recall this too.
Many years ago, and soon after BT bought Plusnet a few of us were invited to trial what Plusnet called the 'RIN' network. Looking back with and with hind sight, we may have been beta testing dedicated WBMC for BTretail. We were issued with BTr IP addresses and as with dedicated you could see all the hops that are normally hidden with shared WBMC... or IPstream as it was back then. The timescale certainly fits .. anyhow I digress slightly, but the interesting thing was the amount of direct peering which cuts out many hops. I gave an example above with the met office. JANET is another, but there are plenty more.. but an interesting one was doing a tracert to someone also on the RIN trial but in the next town to me. My trace was done in about 3 hops total, there was the Manchester RAS (not responding to ICMP), then another IP, then his IP. I could ping him in about 12ms. (ie 6ms to the RAS and 6ms back to him.
Compare this with on shared and how traffic has to go all the way down to London and back up the country through Manchester again and then back on to him, you are talking more like 24ms to ping someone in the same town as me.
So.. as I mentioned in my first post, they could do a lot more direct routing which could cut out the London gateway hops which we are now so used to seeing.
Does the recent introduction of MSE bRAS affect the logical view of "node C"? Does the existence of 50x as many bRAS mean there are many more interconnect nodes, APs and MSILs? Or has the new bRAS placement meant that there is a split in the "21CN backhaul", with one leg from MSAN to bRAS, and another leg between bRAS and interconnect node?
Ive not seen anything documented about this.. my speculation and thats all it is .... to me it seems logical that it doesnt quite do either, but 'split' would be the closest interpretation, because its still in effect 'backhaul' until it reaches the Core node.
They have added a few new nodes but absolutely no where near enough to make all the MSE bRAS core nodes. There always has been some nodes which are core nodes but not interconnects and vice versa. The main benefit of MSE bRAS is authentication occurs closer and sooner. The old style RAS had 2 main functions 1) authentication and 2) routing traffic on to the CORE.
Under the new system
1) Authentication is done nearer as would any policies such as QoS normally carried out at the RAS.
2) 21CN Routing carries on as normal
There is already a structure in place whereby the backhaul traffic passes through a series of switches and routers until it reaches the Core node location. - see the diagram
here.
Apologies if Ive missed an important point when it comes to protocols as thats not my forte and Im far too tired to look it up and think things through properly, so I shall leave someone else to comment on ATM v PPPoE and if that is affected by changes to location.