Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
 
     
   Compare ISP   Rate your ISP
   Glossary   Glossary
 
Please login or register.

Login with username, password and session length
Advanced search  

News:

Pages: 1 [2]

Author Topic: Samknows article on 21CN (old)  (Read 8172 times)

WWWombat

  • Kitizen
  • ****
  • Posts: 1674
Re: Samknows article on 21CN (old)
« Reply #15 on: January 09, 2016, 04:07:55 AM »

Having your PSTN line number connected to local exchange and FTTC coming from the larger exchange has one advantage the phone & broadband won't go down at the same time if there is a fault at one exchange.

You might well find that the small "local exchanges" aren't really local exchanges after all ... and are just remote concentrators, with a voice circuit back to a parent local exchange anyway. And where that is the case, what is the likelihood that the parent for voice services is the same exchange as the fibre head-end?

As for the old plans for 21CN voice...
This article represents what I thought happened a few years ago:
http://www.lightreading.com/ethernet-ip/voip-systems/bt-slows-down-21cn-scraps-converged-service/d/d-id/667448

More recently, we've seen that BT have predicted that everyone will be on a "VoIP"-style service by 2025. They didn't use the term "VoIP", and didn't specify how it would be done in any form. A few business-telecom websites have predicted that ISDN circuits will no longer be available, but with no prediction of analogue PSTN.

BT also indicated the same kind of prediction in the government's "digital comms by 2030" consultation.

I can't decide whether they'll finally use MSAN functionality at the exchange and/or FTTC cabinet, or introduce voice interfaces on ADSL/FTTC router/modems.

Whatever they choose, I see it as the first shots fired in the negotiation to get rid of analogue PSTN, and all the regulation that surrounds it. By 2025, System X will be getting on for 40 years old, and a replacement will need to be in mind.
Logged

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 9106
  • Retd sw dev; A&A; 4 7km ADSL2; IPv6; Firebrick
Re: Samknows article on 21CN (old)
« Reply #16 on: January 09, 2016, 04:48:39 AM »

Apol what is a 'fibre head end'?
Logged

loonylion

  • Reg Member
  • ***
  • Posts: 723
Re: Samknows article on 21CN (old)
« Reply #17 on: January 09, 2016, 11:50:31 AM »

Apol what is a 'fibre head end'?

the exchange at which a fttc/fttp line originates. As stated above, it's not necessarily the same exchange as the telephone line originates from
Logged

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 9106
  • Retd sw dev; A&A; 4 7km ADSL2; IPv6; Firebrick
Re: Samknows article on 21CN (old)
« Reply #18 on: January 09, 2016, 12:33:08 PM »

@loonylion  so does that mean that individual fibre optic cables run all the way from an alien exchange to a local exchange and then oneward straight out to the various cabs? Or does a local exchange demultiplex and distribute to cabs?
Logged

c6em

  • Reg Member
  • ***
  • Posts: 503
Re: Samknows article on 21CN (old)
« Reply #19 on: January 09, 2016, 12:43:04 PM »

@WWWombat

I'd go for doing it in the FTTC cabinet.
Turn the unit into a mini-exchange.

Long term:
Once all the cabs on an exchange have been FTTC'd move everyone compulsorily onto the FTTC service and close down the exchange based ADSL totally.
Make the FTTC cabs voice capable.

Then close down the smaller local exchanges running everything from regional head end exchanges and then sell off the smaller exchanges for housing - or rather 'a house'.

Logged

Weaver

  • Addicted Kitizen
  • *****
  • Posts: 9106
  • Retd sw dev; A&A; 4 7km ADSL2; IPv6; Firebrick
Re: Samknows article on 21CN (old)
« Reply #20 on: January 09, 2016, 12:46:19 PM »

@c6em what about the long lines?
Logged

c6em

  • Reg Member
  • ***
  • Posts: 503
Re: Samknows article on 21CN (old)
« Reply #21 on: January 09, 2016, 01:21:03 PM »

I would hope that if you got rid of ADSL2 totally then the power masks currently in place on the FTTC service to protect existing ADSL2 users from adverse interference from FTTC would no longer be needed.

That means (again hopefully) with the FTTC service now allowed to run unimpeded those on long lines who at the moment cannot get a FTTC service might then be able to get as good as/somewhat better service from an FTTC connection as they were from an exchange based ADSL2 one.

None of this addresses the issue of how to fix the unfixable problem of the individual long lines which occurs everywhere - remote farms and associated cottages in Wiltshire/Berks have the same issue as well as the truly remote/rural areas of the UK.
Logged

licquorice

  • Reg Member
  • ***
  • Posts: 751
Re: Samknows article on 21CN (old)
« Reply #22 on: January 09, 2016, 01:21:38 PM »

Use semaphore  ;D
Logged

WWWombat

  • Kitizen
  • ****
  • Posts: 1674
Re: Samknows article on 21CN (old)
« Reply #23 on: January 09, 2016, 02:28:01 PM »

@loonylion  so does that mean that individual fibre optic cables run all the way from an alien exchange to a local exchange and then oneward straight out to the various cabs? Or does a local exchange demultiplex and distribute to cabs?

The local/child exchange doesn't do any de-multiplexing.

What likely happens (to any individual cabinet) is one of two things:
a) The fibre spine (*) coming outwards from a neighbouring exchange doesn't just stop at the cabinets within its existing boundary. Instead, it continues over, to "subsume" some of the cabinets. Those cabinets will now belong to the parent of the neighbouring exchange, NGA-wise.
b) Fibre cable is run from the head-end exchange (the parent) out to the child; ducting probably already exists, because there probably used to be junction (trunk) cables between the two exchanges. The fibre will reach the "exchange manhole" outside the exchange building, but doesn't necessarily go inside the building. From there, the fibre can be routed onwards to one or more fibre spines outward from the child exchange's manhole.

Don't think of the child exchanges as being limited to just tiny rural exchanges. It happens to much larger exchanges too.

Back in 2011, some of BT's plans leaked into the wild, giving exact coverage plans for postcodes and cabinets. Part of this information also told you the parent-child relationships - and at this early stage in the rollout, only fairly large exchanges were involved.

In my neck of the woods at the time, I noticed the following:
- Brookwood (10,000 lines) and Woking (24,000 lines) were both children of Guildford (30,000 lines)
- Aldershot (25,000 lines), Blackwater (6,000 lines), Camberley (18,000) and Fleet (16,000 lines) were all children of Farnborough (24,000 lines)

That's one child which was larger than its parent!

I'm not sure that the parent-child relationship has to be either neighbouring or even close. Up here, Filey is a child of York - about 60km apart.

(*) - In the access network, a fibre spine is a major route of multi-core fibre (up to 288 fibres, made up of multiple "elements" or "element tubes", each with 12 fibres). The spine ends up being a daisy chain of this fibre, with a number of "aggregation nodes" sited along it. The AGN's are the point where splices can be made, and small fibre cables are routed out to individual FTTC cabinets or (in FTTP areas) out the splitters used in the passive GPON that BT have chosen to use. In future, they'll become the distribution point for fibres for G.fast nodes and FTTPoD, and likely for any leased-line builds too.

In a town, I imagine that there are a number of spine routes; looking at my town's cabinet locations, I can see 8 or 9 quite natural spine routes that roughly follow main roads in 8 different compass points.

In the NGA network, the fibre-spines are equivalent to today's E-side cables, and the aggregation node is equivalent to the PCP cabinet.

This document gives you details of the aggregation node; the introduction might help, but might confuse with the addition of even more TLA's
http://www.te.com/commerce/DocumentDelivery/DDEController?Action=srchrtrv&DocNm=tc-924-sip-uk&DocType=SS&DocLang=English&s_cid=1046

This ISPReview article also mentions some of the build process:
http://www.ispreview.co.uk/index.php/2015/06/interview-how-openreach-is-creating-a-new-fibre-infrastructure-for-wales.html
Logged

WWWombat

  • Kitizen
  • ****
  • Posts: 1674
Re: Samknows article on 21CN (old)
« Reply #24 on: January 09, 2016, 02:43:22 PM »

Once all the cabs on an exchange have been FTTC'd move everyone compulsorily onto the FTTC service and close down the exchange based ADSL totally.
Make the FTTC cabs voice capable.

The only fly in the ointment being that not enough FTTC port capacity has been built (or allowed for) to allow it to take over either the full ADSL role or the full voice role. Almost every FTTC cabinet would need augmenting with a twin. The extra tie pairs needed might also require an extension to the PCP too.

On the other hand - 2025 will have seen a widespread rollout of G.fast and FTTPoD too. An overlay of those two network might provide enough ports for us to not worry about the size of the (by then) ancient FTTC cabinet.

That means (again hopefully) with the FTTC service now allowed to run unimpeded those on long lines who at the moment cannot get a FTTC service might then be able to get as good as/somewhat better service from an FTTC connection as they were from an exchange based ADSL2 one.

This would bring some properties into reach, as would tactical use of vectoring.

An alternative would be to enable ADSL at the cabinet, but I'm not convinced it would give better range once the power masks are removed.

IIRC, the biggest limitation in range on VDSL2 is actually upstream. I'm not sure the power masks affect that - more likely it is things to do with upstream power back off ... which can be removed with vectoring again. But I've seen nothing that studies range impacts on upstream anywhere, so these guesses remain just that...
Logged

NewtronStar

  • Kitizen
  • ****
  • Posts: 4844
Re: Samknows article on 21CN (old)
« Reply #25 on: January 09, 2016, 07:48:06 PM »


You might well find that the small "local exchanges" aren't really local exchanges after all ... and are just remote concentrators, with a voice circuit back to a parent local exchange anyway. And where that is the case, what is the likelihood that the parent for voice services is the same exchange as the fibre head-end?


This is just my opinion into whats going on the local exchange pots go's to my PCP cabinet the larger (head end/parent exchange) and from here the fibre is coming though ducts and then terminates at my FTTC cabinet now this is why we need the tie cable to join the PSTN exchange line with the FTTC  :-\

WWWombat could you try and simplify your posts as sometime it comes across as a technical rant which then leaves me understanding less of what you posted.

sincere apologies if my post is to harsh I like it simple  :)

Logged

WWWombat

  • Kitizen
  • ****
  • Posts: 1674
Re: Samknows article on 21CN (old)
« Reply #26 on: January 10, 2016, 01:26:30 PM »

WWWombat could you try and simplify your posts as sometime it comes across as a technical rant which then leaves me understanding less of what you posted.

sincere apologies if my post is to harsh I like it simple  :)

Oooph. I felt that  :-[

You're quite right, of course. I'm very technical, and can write about that stuff at a very high level. I can also simplify stuff for the audience if needed, but it is indeed hard to do both at the same time. Perhaps I need to do a "TT;DR" (a "too technical" equivalent to the more famous TL;DR) section, or take some lessons from Wikipedia's advice

If anyone wants me to re-state something into a more understandable form, they just need to let me know.

You might well find that the small "local exchanges" aren't really local exchanges after all ... and are just remote concentrators, with a voice circuit back to a parent local exchange anyway. And where that is the case, what is the likelihood that the parent for voice services is the same exchange as the fibre head-end?

Let me revisit this statement, then...

I'm starting from the discussion on the BT architecture for NGA, where some exchanges become a "parent" for NGA fibre services (by having a fibre head-end, containing the optical OLT equipment), leaving many exchanges as "children" without such equipment - the fibre just passes on by.

What I wanted to point out was that there is nothing new in this architecture: that the history of telecoms is full of cases where "child exchanges" are, in reality, a very cut-down model that cannot provide much of a service alone, and are dependent on the existence of a more-complete "parent" exchange.

We recently discussed a model from the Strowger days - the 200-line UAX-13 meant for rural areas. The intention was that this was fully automatic "child" - and that any service requiring an operator (by dialling '0') was passed up to the parent exchange, as would services requiring longer-distance access (by dialling '9'). At full size, this "child" exchange could service 200 subscriber lines, with perhaps 40 lines connecting it to the parent (known as junctions), and perhaps a maximum of 40 parallel calls.

This same "parent-child" arrangement continues into the days of digital telecoms, with the architecture featuring in both the System X and System Y (AXE-10) systems used by BT.

The smaller "child" exchanges are again simplified installations, dependent on the full parent exchange for access to the complete set of services - and, in particular, access to the trunk network. The "child" exchanges are known as "remote concentrators", while the parent exchanges are known as "digital local exchanges".

I know the internals of the AXE-10 better than I do the System X, so I can decipher some of the statistics for their "RSS" (remote subscriber switch), which is used as the System Y "remote concentrator". One "child" RSS can support up to 2,000 subscriber lines, with perhaps 500 lines being used to connect it back to the "parent" 'digital local exchange'. Hence the name "concentrator" - because the unit concentrates 2,000 subscribers down into just 500 connections back to the parent.

A "TT;DR" aside: For those who want serious details on the AXE-10 for "small" applications, there is a lot of detail in a 1990 Ericsson publication. If you take a look and think of that as heavy going, please be assured that it is a reasonable non-technical summary! The real technical details are excruciatingly worse  :-[

For an idea of scale ...
- We, as members of the public, think of BT as having something like 5,000-6,000 exchanges dotted around the country.
- BT actually only has something like 800 "digital local exchanges" that are proper, complete, "parent" exchanges.
- BT also has 7,000 remote concentrators.
- BT has another 150-200 exchanges, whose job is on the trunk network side, without any subscribers at all
(All details from this site, describing the BT telephone network - marked as "current" in 2000, so prior to any 21CN changes)

My post to @NewtronStar ended by pondering whether the new "parent-child" relationships between exchanges for NGA fibre would happen to mirror the existing "parent-child" relationships for voice within the System X/System Y design.

Hope this lays it out a little better...
Logged

NewtronStar

  • Kitizen
  • ****
  • Posts: 4844
Re: Samknows article on 21CN (old)
« Reply #27 on: January 10, 2016, 03:53:44 PM »

That is a nice peice of work WWWombat and very interesting thanks for simplified version  :)
Logged

aesmith

  • Kitizen
  • ****
  • Posts: 1010
Re: Samknows article on 21CN (old)
« Reply #28 on: January 10, 2016, 05:32:49 PM »

I wonder how that parent child relationship applies to AXE10 exchanges where the processor is the other side of the county.  For example I understand that Elgin has it's processor in Aberdeen somewhere, but does that make it a child of one of the Aberdeen exchanges?  Or is the parent child relationship more to do with cable runs, for example where services are provided from a nearby but otherwise unrelated exchange (example when we were served from Lumphanan but had ISDN, the ISDN had a Banchory phone number).
Logged

WWWombat

  • Kitizen
  • ****
  • Posts: 1674
Re: Samknows article on 21CN (old)
« Reply #29 on: January 11, 2016, 03:14:02 PM »

The AXE10 architecture splits the work between a "central processor" (actually two of them, for failover/reliability purposes) and many,many "regional processors". Standard RP's are distributed amongst the hardware on the shelves of the exchange, and perform all the detailed management of that hardware. However, the RSS is controlled by RP's too, but these RP's are located out remotely with the RSS itself. These RP's don't communicate with the CP by the local bus within the exchange, instead communicating over a standard 64Kbps signalling channel, within a standard 2Mbps PCM 30-channel connection (the 30 channels themselves will carry the voice signals between the child and parent).

In this setup, the parent-child relationship is all about control between the CP and its remote RP's, through the cables connecting them. So, yes, it is about the cable run. However, the cable-run doesn't have to be a direct connection, and can run via other sites. Those sites are likely to be children too - more RSS's.

I have to say that I have no idea how far an RSS can sit from the main exchange containing the CP, and was quite surprised to hear of a relationship between Elgin and Aberdeen. That must be 80km direct, or 120km via Peterhead!

I note from the "North of Scotland" fibre-build map visible on this Youtube video, that the existing DWDM route between Aberdeen and Elgin goes via Peterhead, suggesting the 120km figure might be the right one.

I did find another Ericsson publication, this time detailing network planning maps for rural uses of AXE10. Figures 11, 13 and 16 are the useful ones.

In there, Network A has a mesh of RSS's, with the furthest one being about 20km from the main exchange. Network B has the furthest RSS being 95km, but other parts are 120km apart.

It seems the distances can indeed be large!
Logged
Pages: 1 [2]