IP Technical question

[Richard Clayton]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

In article <5024ac90c8ukcrypto@vigay.com>, Paul Vigay
<ukcrypto@vigay.com> writes

>In article <CaUiwtMheCgJFAPm@perry.co.uk>,
>   Roland Perry <lists@internetpolicyagency.com> wrote:
>
>> If I have a high bandwidth consumer service and "unofficially port" it
>> to a different physical line, won't there be the potential for
>> complaints that the speed isn't what was expected?
>
>I think that's quite unlikely because you couldn't port it to a line which
>wasn't using the same ISP 

I think, from my understanding of the mechanisms underneath all of this,
that you should be saying "LLU" here not "ISP" (where one counts a BT
service as one of the LLUs). That is, I believe that you may be able to
use credentials for a different ISP provided that they are a fellow
reseller of the same network service.

However, because of some of the other network structure (as alluded to
by Clive) you may find that it doesn't work -- which I believe is why
no-one offers a service whereby you choose your ISP from hour to hour,
but employ the same piece of copper to connect over.

>- and if you were moving house, you'd presumably
>ask the ISP to setup a new account on the new line - at which point you'd
>be advised as to the capability of that line.

The line will only be connected to a DSLAM if someone appears to be
paying for that, likely the new owner -- or if the previous person did
so, and the cancellation hasn't been fully actioned yet.

Basically, the copper connection to the DSLAM must be paid for, and the
ISP subscription (viz: the RADIUS server saying "OK, you can connect out
to the Internet") is also paid for. Since the paying for the copper
connection is done by paying an ISP and them paying the LLU operator,
there is limited scope for pulling a fast one...

... if all lines were the same speed and all ISPs the same price (which
is true to an order of magnitude) then you can't get a huge economic
advantage from mix-and-match  (it's not like the old days when some ISPs
found that a single dial-up account was paid for and dozens of people
used it simultaneously).

viz: there's no huge economic incentives here (viz although prices and
speeds aren't all the same, there's too little revenue loss for anyone
to care). So the ISP/LLU providers have no interest in fixing up the
system to make the traceability work perfectly.

So law enforcement's piggy-backing onto the system delivers a little
less traceability than you might expect, and if there is no regulatory
pressure to do better, then you cannot expect the traceability to
improve (remember that economics explains what computer science cannot!)

In fact, this neatly illustrates how the Home Office approach in this
area is so fundamentally flawed.

They do not regulate to set out their requirements and leave the ISP to
work out how to do it, which would be efficient and effective. Instead
they pretend to understand the mechanisms (usually from several years
ago) and so they specify precisely which logs are to be recorded and
assume that this will be sufficient, because it might have been once...

They could say "you must be able to translate from an IP address that is
no more than x months old into a customer premises address and tell us
who paid for the service"

Instead they say you must keep:

  * the user ID allocated (this means a unique identifier allocated to
    persons when they subscribe to or register with an internet access
    service or internet communications service).
  * the name and address of the subscriber or registered user to whom an
    Internet Protocol (IP) address or user ID was allocated at the time
    of the communication.
  * the date and time of the log-in to and log-off from the internet
    access service, based on a specified time zone,
  * the IP address, whether dynamic or static, allocated by the internet
    access service provider to the communication, and
  * the user ID of the subscriber or registered user of the internet
    access service;
  * the digital subscriber line (DSL) or other end point of the
    originator of the communication.

which I expect made sense to someone before it was translated into
Spanish and back again by people who'd never done TCP/IP 101 :(

- -- 
richard                                              Richard Clayton

They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety.         Benjamin Franklin

-----BEGIN PGP SIGNATURE-----
Version: PGPsdk version 1.7.1

iQA/AwUBSYBCPJoAxkTY1oPiEQJV/gCfZEaK72N17VxJfG5I6Wy/TC9mlPsAnjyJ
kOq/J6gaIosD9S76CeL3v9RP
=GFAT
-----END PGP SIGNATURE-----