Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 29 Jun 1996 11:22:54 -0400
From:      "Jacob M. Parnas" <jparnas@jparnas.cybercom.net>
To:        Michael Smith <msmith@atrad.adelaide.edu.au>
Cc:        stesin@elvisti.kiev.ua, Kevin_Swanson@blacksmith.com, hardware@freebsd.org, bsdi-users@bsdi.com
Subject:   Re: muliport boards - building a PPP dialup server 
Message-ID:  <199606291523.LAA07860@jparnas.cybercom.net>
In-Reply-To: Your message of Sat, 29 Jun 1996 16:40:43 %2B0930. <199606290710.QAA20610@genesis.atrad.adelaide.edu.au> 

next in thread | previous in thread | raw e-mail | index | archive | help


In message <199606290710.QAA20610@genesis.atrad.adelaide.edu.au>you write:
>Jacob M. Parnas stands accused of saying:
>> 
>> I'm confused.  I thought the 16550 was good up to 115,200 baud, but when
>> ISDN eventually takes over with compression, ~512kbaud will be the norm.
>> I don't know if they can handle that.  I've heard of a 16650 to come out that
>> can do better.  
>
>The 16550 can be clocked safely to several megabits/s.  The problem
>with their implementation in the PC is that they're clocked at
>~1.8MHz, which limits them to 115k2.  The 16650 is a 16550 with a 32-bye
>FIFO.  Neither are suitable for 512KBps; at this speed you have a new
>byte arriving every 16us, or a full FIFO after only 512usec.  This is too
>fast for anything other than a dedicated system.
>
>The best way to do ISDN into a PC is with a $20 ethernet card and a 
>single-MAC ethernet<->ISDN T/A.

Thanks for the information.  But as I said in a recent message, the new TI
chip can go over 900Kbaud/sec.  This isn't so fast.  Its 1/10th the speed
of old ethernet and 1/100 of new 100 Mbit/sec ethernet.

>> Its amazing that one of the slowest protocols on the computer is the
>> one that gives most people the most trouble.  The UARTs seem to be
>> one step behind as do the motherboard/software companies.  I'm of
>> course talking about RS/232 system, which has been a problem with
>> some software at anything over 300 baud at times (even with 9600
>> baud modems operating at 300 baud.  Sun requires a special board at
>> least on my sparc 2.  So there's no problem with a 10mbit/sec
>> ethernet, but 1200baud to 1/20 of the original ethernet speed is
>> even on workstations.
>
>The problem is one of backwards compatability.  There are plenty of
>UARTs around that work, and work well, at higher speeds.  The 8250
>however was an egregious crock when it was first released in the late
>1970's, and has not aged well at all.  None of the better UARTs have
>achieved sufficient acceptance to make any serious headway.

I understand, but it still seems funny that a computer shipped in 1990 or so
(The IBM RT running BSD and with the enhanced advanced processor), couldn't 
put up to date UARTS on the motherboard UARTS or 4 port card uarts, and get
it working fast and reliably.  I think the one on the motherboard had 1 or 2
bytes of buffering.

>> On the positive side, I'm told that CDA is going to port to BSDI
>> next for the Sportster 128 Kbaud (512kbaud with compression) and
>> should have a beta copy in late July.  THey've already done it for
>> Unixware and SCO and possibly Solaris (I think they said they were
>> finishing it up).  Anyway, this is a cheap internal card which you
>> can read all about on http://www.usr.com/.  I've seen it for $319,
>> CDA said that Bell Atlantic and possibly other companies would sell
>> it for $199 if you bought it at the same time as ISDN service.
>
>Of little or no interest on this side of the pond 8)

>> PS.  If I were basing a business on modems, I'd make absolutely
>> certain that they all worked at 512Kbaud, at the same time.  If
>> you've ever experienced
>This is just stupid.  Look at the huge installed base of 14k and 28k
>modems, and the phenomenal cost of ISDN services in most of the world.
>There's no chance that in the short to medium term things are going to
>change; and thus making a serious investment in modern modem
>technology is not unreasonable.

Technology moves pretty fast sometimes.  Just a couple years or so
V.32bis modems or expensive V.FC/V.fast modems were the norm.  Now were
using V.34bis.  Remember how CD's weren't going to replace records for
the most part?  At least on this side of the pond, ISDN is getting
cheaper and cheaper and large graphics files (using netscape) or NFS
may work for some ISDN users depending on the size of the files.  Also,
the quality of network modems should be quite good with ISDN while its
poor on V.34 modems.  I'm not saying its for everyone, but I know my ISP
just spent 15,000 on a server for about 100 ISDN connections.

Even forgetting ISDN, 115,200 baud would be a problem for some UARTs,
which is pretty common these days. (although that's with compression.
I rarely see over 70,000-80,000 baud.)

I think that ISP's have really saturated the market especially with low cost
competition by nationwide companies.

Also, the user base is growing, and many Americans may want to pay a bit
more for their ISDN connection than waiting 10 minutes for a full screen 
24 bit color image to come through via netscape.



The costs are $68 for my line install (cheaper, I think than my analog second 
line)
$25-35/month telco (about same as analog)
about $60 vs $20 for analog for unlimited usage.
$.01/min/channel (biggest problem.  In southern CA, I have a friend who doesn't 
the surcharge per minute on weekends and I think northern CA may be even 
cheaper.
$400 for "modem/terminal adapter" and Unix driver. (may be lower in
   some places.

The ISDN card has the ability to attach an analog modem, phone, fax or whatever
and use one of the channels for free up to 56 kbaud.

Some of the less developed countries may be slower to match US prices, but
I hardly think this solution is stupid for a large section of the population.
Many analog modems cost much more (especially the very early V.fast/V.FC modems.

>> overruns, its not pleasant.  Usually the machine kernel panics and crashes
>> the machine, fairly often, or in clusters.  Not much fun, especially when
>> you have users that have to deal with it, and are justifyably less than
>> pleased at placing their trust in you.

>I have no idea what sort of hardware or software you're running that
>panics or crashes on serial overruns, but I suggest you replace it
>immediately with FreeBSD-based systems, as they don't. 8)

It was an IBM RT running 4.3 BSD with or without the optional 4 port card.
Its no longer a problem and hasn't been since around 1987-8 or so for me.

I prefer to be able to get a contracted support policy, which I don't think
FreeBSD has.  Therefore, I'm going with BSDI.  I'd rather not be down for
a long time because of maintainer of a piece of code is on vacation for 3
weeks.  BSDI has a paid for support contract which requires them to fix
things promptly for not much money.

After being burned by it once, I've been careful since to avoid such problems.

>> | Jacob M. Parnas
>
>-- 
>]] Mike Smith, Software Engineer        msmith@atrad.adelaide.edu.au    [[
>]] Genesis Software                     genesis@atrad.adelaide.edu.au   [[
>]] High-speed data acquisition and      (GSM mobile) 0411-222-496       [[
>]] realtime instrument control          (ph/fax)  +61-8-267-3039        [[
>]] Collector of old Unix hardware.      "Where are your PEZ?" The Tick  [[
>




------------------------------------------------------------------------------
| Jacob M. Parnas                                                            |
| IBM Thomas J. Watson Research Ctr.                                         |
| Internet: jparnas@jparnas.cybercom.net                                     |
------------------------------------------------------------------------------



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199606291523.LAA07860>