[c-nsp] ATM over a Serial Interface question

[Nick Voth]

Yes these end users are ADSL users and in many cases these are very rural
connections where a single T1 will suffice. Often we bond several of them
with IMA to get better capacity, but higher bandwidth links are unavailable.

Thanks,

-Nick Voth


> From: Scott Granados <scott at granados-llc.net>
> Date: Tue, 18 Jan 2011 16:45:44 -0800
> To: Michael Sokolov <msokolov at ivan.harhan.org>
> Cc: "cisco-nsp at puck.nether.net" <cisco-nsp at puck.nether.net>,
> "nvoth at estreet.com" <nvoth at estreet.com>
> Subject: Re: [c-nsp] ATM over a Serial Interface question
> 
> There are many cases when terminating a DSLAM with a single T1 makes sense.
> 
> You assumed that these were ADSL end users, they may not be.  We used this
> exact model in a previous gig to terminate a bunch of management interfaces on
> hardware or console servers.  There's lots of telemetry type applications
> where this setup is more than enough to meet the need.  Also, the DSLAMs I am
> familiar with (samsung, redback and nokia) can terminate DS3 or DS1.
> 
> On Jan 18, 2011, at 4:29 PM, Michael Sokolov wrote:
> 
>> Nick Voth <nvoth at estreet.com> wrote:
>> 
>>> We have a number of channelized DS3's that are terminating on to a 7206VXR
>>> with the standard old PA-MC-T3 cards. We have a need to deliver ATM based
>>> DS1's out to Adtran DSLAMs in the field. Customers will connect to the
>>> Adtran units via ADSL and their local telco loop and their data is then
>>> aggregated by the Adtran and on to the ATM based T1 coming back to us.
>> 
>> Let me see if I've understood you correctly: you are planning on feeding
>> a whole DSLAM with a single T1?  In other words, the aggregated bandwidth
>> usage of all subscribers you are going to serve is going to fit into a
>> single 1.5 Mbps T1?  I thought today's spoiled ADSL users demand a heck
>> of a lot more than that...  (The latter thought is based on my personal
>> experience of the level of ridicule I get for promoting my beloved SDSL
>> which also maxes out at 1.5 Mbps.)
>> 
>>> SO, my question is if we can deliver these circuits to the Adtran units
>>> using the "encapsulation atm-dxi" on each serial interface. I know this
>>> encapsulation is for delivering ATM through an ATM ADSU, but in this
>>> application, does the far end Adtran function as the ADSU or are we missing
>>> something on our end?
>> 
>> I invite someone from Cisco to prove me wrong, but my understanding is
>> that "encapsulation atm-dxi" is another term for what's also known as
>> ATM FUNI:
>> 
>> http://ifctfvax.Harhan.ORG/OpenWAN/l2conv.html
>> 
>> With ATM FUNI aka DXI the line carries HDLC frames which have the AAL5
>> packets encapsulated in them, so the hardware is still HDLC.  True ATM
>> over DS1, OTOH, would mean running actual 53-octet ATM cells over the
>> DS1 bits, which calls for significantly different hardware.  I would
>> expect that a DS1 feeding a DSLAM would need to be true ATM, but I'm not
>> familiar with Adtran DSLAMs specifically.
>> 
>> (The DSLAMs which I do know want to be fed with DS3 or higher, not DS1,
>> and they do want ATM cells over that DS3, not FUNI or DXI.)
>> 
>> HTH,
>> MS
>> _______________________________________________
>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>