asterisk users - Jan 2006 - * point to point t1 solution? / alternatives

This has been an interesting discussion for me (except for the 
sniping).  The last post led me, out of curiosity, to this wiki entry:

http://www.voip-info.org/wiki-Asterisk+TDMoE

I was unaware of this feature, and it looks pretty good.  I've been 
pondering replacing some T1's by leveraging IP capacity but of course 
have run up against the QoS issue.  My idea was different...

I don't have production experience with precisely this type of 
application, but I ask for validation from this list.  Pardon me for 
stating what is undoubtedly obvious to many...

The key to assuring adequate performance in replacing a TDM link with IP 
is to assure that adequate idle time is reserved for voice on the IP 
segment(s) involved in the route.  In this way, latency can be 
stabilized, and if maintained below a certain (arbitrary) threshold, 
performance can be deemed acceptable.

The first step, of course, is to assure that the virtual TDM allocation 
does not exceed the available IP bandwidth (so leave a margin, which is 
huge in the example given).  The next step is to use routers which 
respect the TOS field (however it is used; diffserv/whatever), and 
finally, to assure that no non-VoIP traffic can be injected into the 
path with higher routing priority.

On a point-to-point link, a pair of typical Linux boxes can do all 
this.  Given the original problem, I would place Asterisk boxes at 
either end of the link, and have them blend the ordinary traffic with 
the VoIP traffic (which would probably use IAX to relay calls between 
the T1s), while assuring (enforcing) that VoIP packets are marked as 
highest priority.  There are varied ways of accomplishing this, and a 
good reference which I've used in the past can be found at:

http://www.lartc.org/lartc.html

Additionally, I think one could use the tunneling  techniques described 
in that guide to encapsulate the non-VoIP traffic such that its packets' 
originally marked TOS values are preserved for transit outside the 
segment used for TDM emulation.  In this way, that part of the segment 
bandwidth not required for VoIP would function as a dedicated link, 
allowing other prioritization of traffic such as interactive vs. bulk 
(or even other voice!), with the added advantage that it could use the 
reserved VoIP bandwidth when it is otherwise not required (albeit in the 
case of a T-1 over 10Mb, that's insignificant).

Is this easier or harder than TDMoE as described?  Does the TDMoE shared 
idle bandwidth?  What about stability (I'm thinking of SW releases)?  
What other drawbacks or advantages are there?
>Date: Wed, 25 Jan 2006 23:53:59 -0700
>From: "Damon Estep" <damon@suburbanbroadband.net>
>Subject: [Asterisk-Users] * point to point t1 solution?
>To: "Asterisk Users Mailing List - Non-Commercial Discussion"
>	<asterisk-users@lists.digium.com>
>
>Can anyone point me to a reference or sample config for bypassing a
>nailed up (point to point) t1 between two PBXs with asterisk and a pair
>of t1 cards?
> 
>Right now I have 2 Nortel norstars connected to each other via a leased
>line t1. I also have a solid 10mbps low latency microwave link between
>the 2 sites.
> 
>My goal is to run an asterisk box at each end with a t1 card and
>Ethernet card to act as a TDM<>SIP gateway to bypass the nailed T1 in
a
>relatively dumb configuration, with the goal of migrating off of the
>norstars eventually.
>
> In past situations I would have done this with a pair of Cisco routers
>with T1 interfaces in them but in this case I want to get asterisk into
>the picture as an eventual replacement for the norstars.
>
> 
>  
>

Remember, however that TDMoE is TDMoE, not TDMoIP - it's not routable
(unless you encapsulate it somehow, I guess).

-----Original Message-----
From: asterisk-users-bounces@lists.digium.com
[mailto:asterisk-users-bounces@lists.digium.com]On Behalf Of Bill
Michaelson
Sent: 26 January 2006 14:58
To: asterisk-users@lists.digium.com
Cc: vheether@ltps.org
Subject: Re: [Asterisk-Users] * point to point t1 solution? /
alternatives


This has been an interesting discussion for me (except for the 
sniping).  The last post led me, out of curiosity, to this wiki entry:

http://www.voip-info.org/wiki-Asterisk+TDMoE

I was unaware of this feature, and it looks pretty good.  I've been 
pondering replacing some T1's by leveraging IP capacity but of course 
have run up against the QoS issue.  My idea was different...

I don't have production experience with precisely this type of 
application, but I ask for validation from this list.  Pardon me for 
stating what is undoubtedly obvious to many...

The key to assuring adequate performance in replacing a TDM link with IP 
is to assure that adequate idle time is reserved for voice on the IP 
segment(s) involved in the route.  In this way, latency can be 
stabilized, and if maintained below a certain (arbitrary) threshold, 
performance can be deemed acceptable.

The first step, of course, is to assure that the virtual TDM allocation 
does not exceed the available IP bandwidth (so leave a margin, which is 
huge in the example given).  The next step is to use routers which 
respect the TOS field (however it is used; diffserv/whatever), and 
finally, to assure that no non-VoIP traffic can be injected into the 
path with higher routing priority.

On a point-to-point link, a pair of typical Linux boxes can do all 
this.  Given the original problem, I would place Asterisk boxes at 
either end of the link, and have them blend the ordinary traffic with 
the VoIP traffic (which would probably use IAX to relay calls between 
the T1s), while assuring (enforcing) that VoIP packets are marked as 
highest priority.  There are varied ways of accomplishing this, and a 
good reference which I've used in the past can be found at:

http://www.lartc.org/lartc.html

Additionally, I think one could use the tunneling  techniques described 
in that guide to encapsulate the non-VoIP traffic such that its packets' 
originally marked TOS values are preserved for transit outside the 
segment used for TDM emulation.  In this way, that part of the segment 
bandwidth not required for VoIP would function as a dedicated link, 
allowing other prioritization of traffic such as interactive vs. bulk 
(or even other voice!), with the added advantage that it could use the 
reserved VoIP bandwidth when it is otherwise not required (albeit in the 
case of a T-1 over 10Mb, that's insignificant).

Is this easier or harder than TDMoE as described?  Does the TDMoE shared 
idle bandwidth?  What about stability (I'm thinking of SW releases)?  
What other drawbacks or advantages are there?
>Date: Wed, 25 Jan 2006 23:53:59 -0700
>From: "Damon Estep" <damon@suburbanbroadband.net>
>Subject: [Asterisk-Users] * point to point t1 solution?
>To: "Asterisk Users Mailing List - Non-Commercial Discussion"
>	<asterisk-users@lists.digium.com>
>
>Can anyone point me to a reference or sample config for bypassing a
>nailed up (point to point) t1 between two PBXs with asterisk and a pair
>of t1 cards?
> 
>Right now I have 2 Nortel norstars connected to each other via a leased
>line t1. I also have a solid 10mbps low latency microwave link between
>the 2 sites.
> 
>My goal is to run an asterisk box at each end with a t1 card and
>Ethernet card to act as a TDM<>SIP gateway to bypass the nailed T1 in
a
>relatively dumb configuration, with the goal of migrating off of the
>norstars eventually.
>
> In past situations I would have done this with a pair of Cisco routers
>with T1 interfaces in them but in this case I want to get asterisk into
>the picture as an eventual replacement for the norstars.
>
> 
>  
>
_______________________________________________
--Bandwidth and Colocation provided by Easynews.com --

Asterisk-Users mailing list
To UNSUBSCRIBE or update options visit:
   http://lists.digium.com/mailman/listinfo/asterisk-users

Right - so I will assume this makes it slightly more efficient in that 
respect.  And of course, any solution that uses multiple hops brings in 
a raft of considerations for limiting interference by other data streams 
- the essential QoS question.
>Date: Thu, 26 Jan 2006 15:16:25 -0000
>From: "Steve Langstaff" <steve.langstaff@citel.com>
>
>Remember, however that TDMoE is TDMoE, not TDMoIP - it's not routable
>(unless you encapsulate it somehow, I guess).
>
>  
>

I've seen this discussion before. The conclusion was, it is possible to
route TDMoE through a VPN tunnel depending on the tunnel setup you are using
(bridge + tunnel for example) however the latency would make it useless.
TDMoE is designed for the same network. Unfortuanely I can't find a link for
it, but I remember it distinctly. 

Another, large issue, is that TDMoE uses T1 - style bandwidth constantly
whether it is in use or not. Even if it were possible to route it, and even
if the latency problem was solved, can you imagine your bandwidth surcharge
of ~1.5Mbps constant? 

At the end of the day, emulating TDM through the use of IAX and a well
written dialplan is totally the way to go. 

-----Original Message-----
From: Steve Langstaff [mailto:steve.langstaff@citel.com]
Sent: Thursday, January 26, 2006 8:16 AM
To: Asterisk Users Mailing List - Non-Commercial Discussion
Subject: RE: [Asterisk-Users] * point to point t1 solution? /
alternatives


Remember, however that TDMoE is TDMoE, not TDMoIP - it's not routable
(unless you encapsulate it somehow, I guess).

-----Original Message-----
From: asterisk-users-bounces@lists.digium.com
[mailto:asterisk-users-bounces@lists.digium.com]On Behalf Of Bill
Michaelson
Sent: 26 January 2006 14:58
To: asterisk-users@lists.digium.com
Cc: vheether@ltps.org
Subject: Re: [Asterisk-Users] * point to point t1 solution? /
alternatives


This has been an interesting discussion for me (except for the 
sniping).  The last post led me, out of curiosity, to this wiki entry:

http://www.voip-info.org/wiki-Asterisk+TDMoE

I was unaware of this feature, and it looks pretty good.  I've been 
pondering replacing some T1's by leveraging IP capacity but of course 
have run up against the QoS issue.  My idea was different...

I don't have production experience with precisely this type of 
application, but I ask for validation from this list.  Pardon me for 
stating what is undoubtedly obvious to many...

The key to assuring adequate performance in replacing a TDM link with IP 
is to assure that adequate idle time is reserved for voice on the IP 
segment(s) involved in the route.  In this way, latency can be 
stabilized, and if maintained below a certain (arbitrary) threshold, 
performance can be deemed acceptable.

The first step, of course, is to assure that the virtual TDM allocation 
does not exceed the available IP bandwidth (so leave a margin, which is 
huge in the example given).  The next step is to use routers which 
respect the TOS field (however it is used; diffserv/whatever), and 
finally, to assure that no non-VoIP traffic can be injected into the 
path with higher routing priority.

On a point-to-point link, a pair of typical Linux boxes can do all 
this.  Given the original problem, I would place Asterisk boxes at 
either end of the link, and have them blend the ordinary traffic with 
the VoIP traffic (which would probably use IAX to relay calls between 
the T1s), while assuring (enforcing) that VoIP packets are marked as 
highest priority.  There are varied ways of accomplishing this, and a 
good reference which I've used in the past can be found at:

http://www.lartc.org/lartc.html

Additionally, I think one could use the tunneling  techniques described 
in that guide to encapsulate the non-VoIP traffic such that its packets' 
originally marked TOS values are preserved for transit outside the 
segment used for TDM emulation.  In this way, that part of the segment 
bandwidth not required for VoIP would function as a dedicated link, 
allowing other prioritization of traffic such as interactive vs. bulk 
(or even other voice!), with the added advantage that it could use the 
reserved VoIP bandwidth when it is otherwise not required (albeit in the 
case of a T-1 over 10Mb, that's insignificant).

Is this easier or harder than TDMoE as described?  Does the TDMoE shared 
idle bandwidth?  What about stability (I'm thinking of SW releases)?  
What other drawbacks or advantages are there?
>Date: Wed, 25 Jan 2006 23:53:59 -0700
>From: "Damon Estep" <damon@suburbanbroadband.net>
>Subject: [Asterisk-Users] * point to point t1 solution?
>To: "Asterisk Users Mailing List - Non-Commercial Discussion"
>	<asterisk-users@lists.digium.com>
>
>Can anyone point me to a reference or sample config for bypassing a
>nailed up (point to point) t1 between two PBXs with asterisk and a pair
>of t1 cards?
> 
>Right now I have 2 Nortel norstars connected to each other via a leased
>line t1. I also have a solid 10mbps low latency microwave link between
>the 2 sites.
> 
>My goal is to run an asterisk box at each end with a t1 card and
>Ethernet card to act as a TDM<>SIP gateway to bypass the nailed T1 in
a
>relatively dumb configuration, with the goal of migrating off of the
>norstars eventually.
>
> In past situations I would have done this with a pair of Cisco routers
>with T1 interfaces in them but in this case I want to get asterisk into
>the picture as an eventual replacement for the norstars.
>
> 
>  
>
_______________________________________________
--Bandwidth and Colocation provided by Easynews.com --

Asterisk-Users mailing list
To UNSUBSCRIBE or update options visit:
   http://lists.digium.com/mailman/listinfo/asterisk-users
_______________________________________________
--Bandwidth and Colocation provided by Easynews.com --

Asterisk-Users mailing list
To UNSUBSCRIBE or update options visit:
   http://lists.digium.com/mailman/listinfo/asterisk-users

She ain't cheap, but this'll work:

http://www.blackboxcanada.com/Catalog/Detail.aspx?cid=381&mid=4291

It's TDMoIP so 2 T1 boxes tied together should work like this:

T1--TDMXX card--Asterisk--TDMXX card--Voice Mux--Broadband--Voice Mux--TDMXX
card --Asterisk

at about $7K Cdn it'd be worthwhile to rewrite a dialplan to use IAX
instead.

I can appreciate the desire to avoid reconfiguring existing hardware, 
but that is part and parcel of what we are discussing: reconfiguring 
hardware.  Without further specification, it has no bearing on how to 
preserve application behavior, which is what we are trying to accomplish 
with this discussion.

I don't wish to second-guess your analysis of the business requirements 
- you are the authority - but your initial post expressed a desire to 
move toward an Asterisk configuration as one of your goals.  Toward that 
end, development of an appropriate dialing plan ultimately must happen, 
and I would think if done properly, would not change dialing patterns or 
extension numbering unless this is what you desired.

I must agree that fax and modem performance is problematic, but here 
again, this would be an issue anyway when you transition completely to 
Asterisk, as you implied about your long-term plan.  So perhaps now is 
the time to address this matter.

Are you sure you really want to do this at all?
>Date: Thu, 26 Jan 2006 08:52:04 -0700
>From: "Damon Estep" <damon@suburbanbroadband.net>
>  
>
what are the application implications of this? In other words, what 
system behavioral changes will your users experience in the various 
scenarios (pure circuit emulation vs. relay via IAX or similar)?
>circuit emulation will;
> 
>1. eliminate the need to reconfigure the exisitng hardware.
>2. improve the chances that fax and analog modem devices will still work.
>3. NOT change any dialing patterns or extensons numbering.
> 
>there are other, but they are less significant
>  
>
>>
>My goal is to run an asterisk box at each end ... with the goal of migrating
off of the
>>norstars eventually.
>  
>

Bill,

The question has been answered; Asterisk is NOT capable of providing a
clear PTP TDM path between two boxes configured with t1 cards and any
type of trunking.

I am aware of my other options, and will move forward with the knowledge
that a solution we have implemented with back to back Cisco routers in
the past is not possible with asterisk.

I do intend to move forward with asterisk in this particular
environment, but I wish to integrate asterisk by placing it between the
telco PRI and the primary Nortel switch, not between the primary switch
and a branch office on a T1 tie trunk.

I will leave the leased PTP t1 in place for now since it cannot be
dropped without significant additional work for a temporary situation.

BTW, I did learn that the RAD IPMux-11 will do exactly what we need, but
cost is about $1500/pair, so the break even on a $300/mo leased line is
5 months, in which time we are not planning to need the T1 anymore. With
that said, it is still about $6k less than the Cisco solution.

Thanks for your support on this.

> -----Original Message-----
> From: asterisk-users-bounces@lists.digium.com [mailto:asterisk-users-
> bounces@lists.digium.com] On Behalf Of Bill Michaelson
> Sent: Thursday, January 26, 2006 12:29 PM
> To: asterisk-users@lists.digium.com
> Subject: [Asterisk-Users] Re: * point to point t1 solution? /
alternatives
> 
> I can appreciate the desire to avoid reconfiguring existing hardware,
> but that is part and parcel of what we are discussing: reconfiguring
> hardware.  Without further specification, it has no bearing on how to
> preserve application behavior, which is what we are trying to
accomplish
> with this discussion.
> 
> I don't wish to second-guess your analysis of the business
requirements
> - you are the authority - but your initial post expressed a desire to
> move toward an Asterisk configuration as one of your goals.  Toward
that
> end, development of an appropriate dialing plan ultimately must
happen,
> and I would think if done properly, would not change dialing patterns
or
> extension numbering unless this is what you desired.
> 
> I must agree that fax and modem performance is problematic, but here
> again, this would be an issue anyway when you transition completely to
> Asterisk, as you implied about your long-term plan.  So perhaps now is
> the time to address this matter.
> 
> Are you sure you really want to do this at all?
> 
> >Date: Thu, 26 Jan 2006 08:52:04 -0700
> >From: "Damon Estep" <damon@suburbanbroadband.net>
> >
> >
> what are the application implications of this? In other words, what
> system behavioral changes will your users experience in the various
> scenarios (pure circuit emulation vs. relay via IAX or similar)?
> 
> >circuit emulation will;
> >
> >1. eliminate the need to reconfigure the exisitng hardware.
> >2. improve the chances that fax and analog modem devices will still
work.
> >3. NOT change any dialing patterns or extensons numbering.
> >
> >there are other, but they are less significant
> >
> >
> >>
> >My goal is to run an asterisk box at each end ... with the goal of
> migrating off of the
> >>norstars eventually.
> >
> >
> 
> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
> 
> Asterisk-Users mailing list
> To UNSUBSCRIBE or update options visit:
>    http://lists.digium.com/mailman/listinfo/asterisk-users