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

Re: (ASCEND) Analog Multilink PPP



Jeff,

I have no problem splitting this into two threads. In fact, I'll be happy
to take this conversation offline if that's more appropriate. I'm not sure
how many people are following the dialer conversation vs. the other.

I still don't necessarily understand the comment regarding the inability to
use different vendors modems for an analog MP connection. The MP process is
the process that worries about sequencing and fragmentation/reassembly,
correct? So if one link in the bundle is slower for whatever reason,
shouldn't the MP process deal with this? If I'm missing something here,
please correct me.

Charlie Clemmer

At 12:34 PM 5/23/97 -0700, Jeff_Powell@ascend.com wrote...
>
>This thread has wandered, so I hope we can split it in two.
>
>When I said "I'm not aware of any other....blah blah, etc. etc." I was
>referring to PC based dialers [the title of the thread is "Analog Multilink
>PPP and RAS" after all.  So let's divide this into two discussions:
>
>1.  Asynch MPP [Specifically Windows based PC's dialing into an access
>router/server with more than one modem]
>2. All other MP/MPP types [UNIX machines, ISDN TA's/routers, router to
>router, etc. etc. etc.]
>
>Sound fair?
>
>Now, as to #1.  Microsoft release multiline RAS in NT 4.0 [there's a patch
>for Win 95 as well, part of the ISDN connectivity kit].  Choosing "multiple
>lines" in the "modem properties" dialog in RAS [dial up networking] allows
>you to use two modems to dial into a RAS server.
>
>The AMPP specs don't care about what kind of modems, serial ports, etc.
>etc. that you have, but let's think it through for a minute.  The hardest
>job an MP server has is to synch up the incoming calls.  This is easy in
>the digital world, but in the analog world, it gets quite tricky.  Ascend
>has had support for Analog MP in the MAX for awhile, but the only clients
>that could do it were UNIX machines and some remote access servers.  When
>Microsoft came out with its offering, it didn't work with the Max.
>Basically there was too much slop on the client side of things and the Max
>couldn't compensate [a Pentium box running NT4.0 is still not a REAL
>workstation...].  Since this was a new offering from Microsoft, the
>reworking of MP code in the Max to make it more aggressive [and lenient at
>the same time...] in synching up the calls was put in an "i" release so it
>could get beat up in the real world before going prime-time.
>
>As to the client side, the closer in functionality the serial ports and the
>modems are, the better it works.   The phone lines are not as important.
>I've tested this with one "normal" POTS line and one going through a POTS
>jack on a P25.  On the line going through the P25, I would get 33.6k and on
>the POTS line I would get 31.2k and the lines synched up just fine.  The
>key was that I used a third party 115k serial board [as Com2 and 3] and two
>Courier V.Everything modems [same firmware, etc.].  Any attempts to get a
>session with two different modems or different serial ports failed.
>
>Reports from the field have been positive so far, both in the support for
>Multiline RAS and that other aspects of MP didn't break, so we'll probably
>see this in 5.1xxxx.
>
>Anyway, I hope that clears that up...
>
>
>
>
>
>cclemmer@baynetworks.com on 05/23/97 11:35:46 AM
>
>To:   matt@ascend.com
>cc:   Ascend-users@bungi.com (bcc: Jeff Powell/Ascend/US)
>Subject:  Re: (ASCEND) Analog Multilink PPP and RAS
>
>
>
>
>
>Matt,
>What you describe is what I understood MP to be. The question that comes up
>because of this then, is why would Livingston and maybe Ascend not support
>MP over all types of links?
>I keep seeing mixed responses as to whether the Ascend box does support MP
>over any kind of link, while I saw a response from a Livingston guy who
>indicated that their box only supports MP on ISDN calls.
>Knowing that MP should not be related to the physical call type, why would
>this limitation exist?
>Can we get a definitive answer from Ascend as to whether or not the Ascend
>boxes share this limitation? Again, I would expect that it is supported,
>for the reasons you mentioned.
>Charlie Clemmer
>At 08:41 AM 5/23/97 -0700, Matt Holdrege wrote:
>>Folks, MP operates at a layer above the physical connection or media. You
>>can combine any types of calls in any increment. I've seen NT call the Max
>>with one analog call bundled with one ISDN call. Or two analog calls
>>together. I've seen the Max call a Bay router with 23 channels of MP.
>>
>>All MP does is take (in one case) the endpoint discriminator and compare
>it
>>with any other call it has, be it analog or MP or whatever. If there is a
>>match, it bundles the two calls together. I haven't tested it on the Max
>>5.0ap branch, but by the reports here I might assume that the Max code
>>didn't check ED on analog calls. But the reports say that it does in the i
>>branch.
>++ Ascend Users Mailing List ++
>To unsubscribe:     send unsubscribe to ascend-users-request@bungi.com
>To get FAQ'd:  <http://www.shore.net/~dreaming/ascend-faq>
>or        <ftp://ftp.shore.net/members/dreaming/ascend-faq.txt>
>
>
>
>
>
>
>
++ Ascend Users Mailing List ++
To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd:	<http://www.shore.net/~dreaming/ascend-faq>
or		<ftp://ftp.shore.net/members/dreaming/ascend-faq.txt>