Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: (ASCEND) shared D-Channels on PRI

At 10:51 AM 10/14/98 , Phillip Vandry wrote:
>> Related to this, I had a 3com vendor tell me yesterday that NFAS was a 
>> BAD thing, that above 1 PRI's worth of ISDN calls, that performance
>> would drop off, both for call setup and throughput.  He also claimed
>> that this was true for all equipment including Ascend MAXs.
>
>BS.

I didn't want to go so far, but ... ;)

>(I wonder if it is Lucent who told you this? - They don't support NFAS)

Actually they do since 3.8 ComOS, and they *claim* to have multi-chassis
NFAS too - either in 3.8 or in BETA....

>There is plenty of bandwidth in a 64K channel to handle the call load on
>massive amounts of B channels. I would imagine that even if a large
>number of them were setting up or tearing down calls, it would still be
>enough.

When I worked for AT&T (now Lucent) we did some testing with SS7 trunks
where we had the equivalent of NFAS (I think it was still called NFAS
actually - it's been about 10 years :) and did quite some testing of
a single 64K D-channel 'supporting' tens of spans (I think our test LAB
maxed out at 16 E-1s : E1 has 30 channels to play with!).

>> My first thought was to politely tell him how wrong he was, and then
>> realized that I didn't have enough real world data to form an opinion.
>
>I don't know how big an ISDN conversation is to set up a call, assuming
>it comes with CallerID information and some other goodies, but let's
>say it's about 100 bytes.
>
>100 bytes * 80 calls per second = 8000 Kilobytes per second = 64Kbps

This would assume that they were all at the "SETUP" stage simultaneously.
The CONNect and CONNect ACK messages are much smaller, as are the
ALERTing and CALL PROCeeding messages.

>Do you have 80 calls per second coming in?
>
>Now I _have_ seen some systems that have that kind of a call load or
>more on just 3-4 PRIs, but it's not NASes that are connected to those
>lines.
>
>> Also are these values effected by wheither or not the PRIs are 
>> terminated in 1 chassis or multiple chassis?  Is there an effect 
>> that can be tied to type of calls, Modem vs. ISDN?
>
>Call setup *should* be identical for modem/ISDN.

With the obvious differences in content (speech vs unrestricted digital).

>I *do* imagine, as I said in another message, that terminating the PRIs
>on different boxes is something I wouldn't do. I hypothesize that it
>would incur delays. Not to mention what would happen if you lose a Max
>or have bad collisions on your LAN.

NFAS is NOT currently supported across chassis anyway.


Kevin Smith			(kevin@ascend.com)
Ascend Communications...
	...where Network Solutions never end.

++ Ascend Users Mailing List ++
To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd:	<http://www.nealis.net/ascend/faq>
Follow-Ups: References: