Home page logo
/

nanog logo nanog mailing list archives

Re: DC fiber cut: field report
From: "Sean M. Doran" <smd () clock org>
Date: Tue, 22 Jul 1997 04:24:09 -0700

| The impression I get is that the protection circuit is intended to 
| cover failures that take out a single circuit.

When talking about "protection circuits", bear in mind that
this is all legacy PDH stuff which hopefully soon will be dead.

An STM-1, for example, can be made fault-resistant by virtue of
SDH's ring architechture.  Any PDH payload encapsulated in the
STM-1 consequently will be immunized against backhoe fade, failed
optics and the like.  Obviously a failure that SDH's mechanisms
can't prevent will require any encapsulated traffic to be rerouted
through non-SDH means.  

With IP, this is simple: you see a circuit failure, and change your
routing tables in the normal way.  With many VC-based networks you
tear down all sorts of VCs and try to build them over other paths,
updating your switching tables in the normal way.  Finally, with
things that are reliant on PDH protection, you wait for them to
fall over onto protection circuits, which are laid out in different
STM-1s, ideally along completely different physical paths.

In other words, in the event of a failure that SDH cannot cope with,
you either overscribe other in-use capacity or you make use of 
stand-by capacity that is otherwise completely unused.

Completely unused capacity is expensive, and you already have
quite a bit of it if you build SDH rings (half, in fact).  Adding
more unused capacity to support unlikely failures is unattractive.
However, most failures that SDH cannot cope with tend to be very rare.
I would be surprised, for example, if the fibre cuts in the U.S.
affected SONET-ring-carried traffic, and I would be surprised
if a PDH protection circuit scheme would provide much faster
recovery in the cases where SDH/SONET failed.

Of course, if the lines that failed were, in fact, not PDH, 
I'd appreciate a private word about this. :)

| For example, if you 
| had 10 fibers in a system and a repeater or laser died, then only 
| one of your fibers stops working.  Fixing that with a protection 
| circuit adds 1/10 to the system cost.

Um, huh?  If you have a fibre stop working then all the data
traversing that fibre stops.  If you have an SDH ring, you don't
really care, because along the segment between the two MUXes at
either eind, you have a second fibre that likely remains working.
If you have protection circuits you have to deal with DAXCs and
the like and you have to have the protection circuits provisioned
along another fibre and standing idle.

PDH protection switching is slow and ugly, but it allows 
fall-back connectivity to be oversold or not provisioned in
the first place.  That is, you can sell people "cheapo" circuits
which have no backup, and even more, circuits which will fail
when an unrelated circuit used by a customer buying an expensive
circuit fails.

|  (Maybe less if you have spare 
| fibers in the ground.)  Fixing it with SONET ring technology roughly 
| doubles the system cost. 

If you have spare fibres in the ground, light them up to
protect your primary SDH paths.  Even if you use another pair
of fibres runnning through the same conduit, you do protect
against failed optical and electrical equipment (repeaters,
lasers, fibre tips, people unplugging the wrong thing, and so
on).  You don't need an additional MUX, you don't need extra DAXCs,
and you don't need to scurry around reprogramming lots of equipment
when someone wrecks your fibre.

        Sean.


  By Date           By Thread  

Current thread:
[ Nmap | Sec Tools | Mailing Lists | Site News | About/Contact | Advertising | Privacy ]
AlienVault