asterisk users - Jun 2003 - Clock Sync

There is a little discussion going on comp.dcom.voice-over-ip and I could not
answer these two questions... The author has sorta clarified his two questions,
but they still mean little to me.
> 6) A redundant configuration that can synchronize on, and share one, two,
> and more network clocking signals.  Plus synchronize on a independent
> stratum 3 or greater clock source.
> 8) A redundant configuration that can synchronize on, and share one, two,
>  and more network clocking signals.
> For Question 6, and 8, I was wanting to know how Asterisk handles the
> passthrough of network (T1) clock synchronization from one T1
> interface to another.  This would allow an Asterisk systems to feed
> clock to downstream systems.  Avoiding  frame slips across the voice
> network, where Asterisk is the front end to it all (connected to it
> PSTN), and connecting to other systems via T1.  I am aiming to find
> that kind a mixed network answer.

Anyone know what this guy is talking about?


Jeremy McNamara

On Wednesday, June 04, 2003 1:25 AM, Jeremy McNamara 
[SMTP:jj@indie.org] wrote:
> There is a little discussion going on comp.dcom.voice-over-ip and I
> could not answer these two questions... The author has sorta
> clarified his two questions, but they still mean little to me.
>
>> 6) A redundant configuration that can synchronize on, and share one,
>> two,
>> and more network clocking signals.  Plus synchronize on a
>> independent
>> stratum 3 or greater clock source.
Asterisk can share 1,2,3 or 4 network clocking via incoming T1s. 
However it can not time to an external  stratum 3 or greater clock 
source.
>
>> 8) A redundant configuration that can synchronize on, and share one,
>> two,
>>  and more network clocking signals.
>
>> For Question 6, and 8, I was wanting to know how Asterisk handles
>> the
>> passthrough of network (T1) clock synchronization from one T1
>> interface to another.  This would allow an Asterisk systems to feed
>> clock to downstream systems.  Avoiding  frame slips across the
>> voice
yes * can do pass through timing. Just select 1 of the T1s to be the 
clock source and all others will use this same timing.
>> network, where Asterisk is the front end to it all (connected to it
>> PSTN), and connecting to other systems via T1.  I am aiming to find
>> that kind a mixed network answer.
>
>
> Anyone know what this guy is talking about?
When I was looking at timing before this is the conclusion that I have 
come to.

The T400P card has an internal clock that all four T1s of that card 
will be timed off of. This internal clock can be free running (not 
referenced to any other clock) or reference to another clock (the clock 
on the other end of a T1 line). This is set up in Zaptel.conf
span=<span num>,<timing>,<line build
out>,<framing>,<coding>

Important points.
  1. All T1s of a T400P run off of the same internal clock.
  2. The internal clock can be free running - not reference to any 
other clock.
  3. The only other reference the clock can use is timing from an 
incoming T1. If the primary clocking T1 is down then it can be set to 
use a secondary T1 to derive timing (or 3rd or 4th)
  4. There is no way to time off of an external clock be it stratum 3 
or 2 or whatever (unless it is part of a T1 thereby burning up a T1 
port in the T400P card).
  5. If the system has 2 T400P cards the clocks will not be timed to 
each other (unless each has an incoming T1 with the same timing.)
>
>
> Jeremy McNamara
>
Don Pobanz

Don Pobanz wrote:
>When I was looking at timing before this is the conclusion that I have 
>come to.
>
>The T400P card has an internal clock that all four T1s of that card 
>will be timed off of. This internal clock can be free running (not 
>referenced to any other clock) or reference to another clock (the clock 
>on the other end of a T1 line). This is set up in Zaptel.conf
>span=<span num>,<timing>,<line build
out>,<framing>,<coding>
>
>Important points.
>  1. All T1s of a T400P run off of the same internal clock.
>  2. The internal clock can be free running - not reference to any 
>other clock.
>
You can only use this for a single card within a private network, 
unconnected to the public network. In that case you can make that one 
card free run, and it becomes the network master. All the others must 
slave to it.
>  3. The only other reference the clock can use is timing from an 
>incoming T1. If the primary clocking T1 is down then it can be set to 
>use a secondary T1 to derive timing (or 3rd or 4th)
>
This is what 99% of people need. The public network will not sync to 
you. You must sync to them. Public exchanges actually use rhubidium 
atomic clocks, so their clock is darned accurate (just like the line 
scan rate of your TV signal, and for similar reasons).
>  4. There is no way to time off of an external clock be it stratum 3 
>or 2 or whatever (unless it is part of a T1 thereby burning up a T1 
>port in the T400P card).
>
Stratum 3 is a term I normally associate with NTP. This is a 
macroscopically accurate clock. A clock accurate at the cycle level is 
needed to keep a T1 flowing right.
>  5. If the system has 2 T400P cards the clocks will not be timed to 
>each other (unless each has an incoming T1 with the same timing.)
>  
>
If you are smart this will probably be the case. Make one of the T1s on 
each card a line from the public network, and use 1 to 3 of the others 
for private network lines. All those T1s comming in from the public 
network, even from different carriers, will be in isochronous, so you 
will never get slips this way. It would have been nice if Mark had 
provided a link wire facility between the cards, to keep them all in 
step under all conditions, but life is never perfect. :-\

Regards,
Steve