North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Re: T1 short-haul vs. long-haul - jack terminology

  • From: frank
  • Date: Sat Jul 24 12:14:56 2004

On the matter of the type of cabling to be used between the Telco Demarc and the 
CPE, I have found this to be one of the most shrouded of all areas in telecom 
standards. The jabber and deliberations that have taken place over this issue 
border on folk lore and hijinx, and could fill a plant man's notebook in no time 
flat.

Prior to metallic T1s being made available to end users by the incumbent carriers 
(this is going back some tweny years or so), the carrier itself extended cabling 
to the point of the customer's CPE/DTE. Reportedly (and in fact, most of the 
time), they used individually shielded pairs, actually comprising two separate 
cables - one for each direction. 

The standard that was used in the latter case was the ancestor of today's ANSI 
401.1 and it applied to the "telco's" EMI/RFI compliance responsibilities, not the 
customer's. 

Later on, the EIA/TIA drew up a set of commercial building wiring standards in the 
way of TIA568, but neglected to include the T1 niche in its writings, relegating 
it to an orphan state. Consequently, there has been much confusion amongst 
enterprise IT staff and Telco guidelines, not to mention the list of different 
recommendations that manufacturers of CSUs and DSUs place in their glossies. 

I've seen where STP (shielded twisted-pair cabling) purists have succeeded in 
having shielded cabling used, only to screw it up by mis-applying the necessary 
grounding connections causing more problems than they solved. 

And I've seen where wall fields of T1 demarcs in an MDF room have successfully 
been deployed using nothing more than Cat3 UTP between spaces that were thirty 
stories apart. Through all of this, the one area that is often overlooked is the 
area where vendors of network hardware stipulate what type of cabling they will 
support and will _not_ support, with pair shielding used as a determining factor.

About two years ago, after a client successfully deploying several shelves of T1s 
for most normal enterprise purposes, they found that one certain VoIP gateway was 
encountering excessive errors. After all self tests came up negative, as did an 
end to end test on the T1, and after swapping out those T1s and their respective 
patch cables, the problem persisted. 

At one point the vendor's rep pointed to the fine print in the operator's manual. 
It stipulated that the only cabling that would be supported between the gateway 
and the telco demarc was individually shielded twisted pairs. When it was 
discovered that the client had chosen to use UTP patch cords to extend their T1 
demarcs some fifteen or so feet, the vendor would not even begin to diagnose the 
problem further in depth until the client replaced those cables with shielded 
ones. 

Once replaced, the problem appeared to subside as evidenced by a lower incidence 
of errored seconds initially (but still unacceptable from my view), only to 
resurface at a later time to the same state of impairment that it was before. 

But the salient point here isn't whether or not the problem was solved, as much as 
it is the significance of abiding by vendor warranty conditions (if you've agreed 
to them initially, even tacitly, in your contracts and SLAs) when selecting a T1 
cabling solution. 

If anyone on this list has a more authoritative source that stipulates the type of 
cabling to be used for the section in question (between the demarc and the CPE), 
perhaps something that has been written more recently than the experience I cited 
above, it would be appreciated.

[email protected]

----------


On Fri, 23 Jul 2004 23:52 , 'Forrest W. Christian' <[email protected]> sent:

>
>On Fri, 23 Jul 2004, Christopher Woodfield wrote:
>
>> OK, from my reading in Newton's Telecom Dictionary, it appears that NIU
>> is a generic term for "whatever the customer plugs their cable into",
>> be it a powered or a dumb device. Mea culpa.
>...
>> "...installed on the premises as a semi-intelligent demarcation
>> point,
>> the smart jack is completely passive until activated remotely by a
>> digital code, typically something like 'FACILITY 2', sent down the T-1.
>> This code activates a relay [that loops the circuit]."
>>
>> That may not accurately define the Adtran and Westell devices that are
>> pictured (they appear to have additional features beyond this), but
>> it's a good guess they provide the remote loopback function described
>> above in addition to the monitor points and management console port. I
>> also doubt that the Hyperedge unit pictured does so, although I can't
>> seem to find any online documentation on the unit (it is, as you
>> described it, a 'glorified patch panel'). Feel free to correct me.
>
>In Qwest land, NIU, Smart Jack, and Demarc (unless "extended") are all in
>the same physical rack.
>
>When you get a T1, qwest installs an appropriately sized shelf. This
>shelf holds the adtran and westell devices shown in earlier posts. For
>example, we have one site with quite a few T1's, which they installed a
>rack like the one pictured at:
>
> http://www.westell.com/images/osp/dsawm214.jpg
>
>Note the RJ45's on the bottom. These are the demarc point for the
>circuit. Older ones have RJ45's on the right side and the cards are
>thicker - a lot thicker.
>
>When qwest says "insert a loopback plug at the smartjack" or "unplug from
>the smartjack" or whatever, they mean this device.
>
>Qwest can loop or unloop and do other tests to this device. On the newer
>HDSL cards, they can also plug a laptop in to get performance data, and I
>believe they can also get this data from the CO end.
>
>Also of note, I haven't seen qwest deploy anything but HDSL2 cards for
>quite a while. This basically means a full duplex, full-speed T1 over a
>single pair of copper with a quarter of the repeaters (12K wire feet
>without a repeater).
>
> -forrest 



On Fri, 23 Jul 2004 23:52 , 'Forrest W. Christian' <[email protected]> sent:

>
>On Fri, 23 Jul 2004, Christopher Woodfield wrote:
>
>> OK, from my reading in Newton's Telecom Dictionary, it appears that NIU
>> is a generic term for "whatever the customer plugs their cable into",
>> be it a powered or a dumb device. Mea culpa.
>...
>> "...installed on the premises as a semi-intelligent demarcation
>> point,
>> the smart jack is completely passive until activated remotely by a
>> digital code, typically something like 'FACILITY 2', sent down the T-1.
>> This code activates a relay [that loops the circuit]."
>>
>> That may not accurately define the Adtran and Westell devices that are
>> pictured (they appear to have additional features beyond this), but
>> it's a good guess they provide the remote loopback function described
>> above in addition to the monitor points and management console port. I
>> also doubt that the Hyperedge unit pictured does so, although I can't
>> seem to find any online documentation on the unit  (it is, as you
>> described it, a 'glorified patch panel'). Feel free to correct me.
>
>In Qwest land, NIU, Smart Jack, and Demarc (unless "extended") are all in
>the same physical rack.
>
>When you get a T1, qwest installs an appropriately sized shelf.  This
>shelf holds the adtran and westell devices shown in earlier posts.  For
>example, we have one site with quite a few T1's, which they installed a
>rack like the one pictured at:
>
>  http://www.westell.com/images/osp/dsawm214.jpg
>
>Note the RJ45's on the bottom.  These are the demarc point for the
>circuit.  Older ones have RJ45's on the right side and the cards are
>thicker - a lot thicker.
>
>When qwest says "insert a loopback plug at the smartjack" or "unplug from
>the smartjack" or whatever, they mean this device.
>
>Qwest can loop or unloop and do other tests to this device.  On the newer
>HDSL cards, they can also plug a laptop in to get performance data, and I
>believe they can also get this data from the CO end.
>
>Also of note, I haven't seen qwest deploy anything but HDSL2 cards for
>quite a while.   This basically means a full duplex, full-speed T1 over a
>single pair of copper with a quarter of the repeaters (12K wire feet
>without a repeater).
>
> -forrest