Conference 7.286::digital

    See notes on the Internet Tunnel in the LJSRV2::INTERNET_TOOLS
    notesfile.

Some Digital dial-in services provide SLIP/PPP access.  Contact
your local telecommunications support. 

Wes

Digital has been experimenting, in the UK at least, with ISDN and high speed 
leased line access to the EasyNet for people who are `heavy' users of the 
network.  You may want to put this idea to someone.

Chris.

    1.	Get a regular ACB account.
    
    2.	Find someone with an OSF (sorry, I mean Digital Unix) box.
    
    3.	Get a copy of SLIRP to run on the OSF box (put "SLIRP" into Alta
        Vista to find a copy, ftp://blitzen.canberra.edu.au/slirp/
        slirp-1.0c.tar.gz is one place to look).
    
    If you have Windoze95 with dial-up networking (get the plus pack with
    the scripting tool), you will be able to run TELNET, NETSCAPE and even
    connect to network drives from home !  Even eXcursion and Teamlinks work !
    You don't even need your own internet address; SLIP works by using the
    address of the box you run SLIRP on.

ISDN?  In New England?  You must be kidding!

It's practically impossible to get NYNEX to sell it even in those few places
where it's available.

And non-ISDN (28.8 bps) calls from home can be untimed/uncharged, whereas
ISDN in New England _always_ costs about 1.5 cents/minute.

/john

    We've put in ISDN lines for our employees in my group.  We haven't
    bought them home pcs, however, they had to either have those
    already or pony up for them (only if they wanted ISDN).  PacBell
    offers various group discount plans for businesses that we're
    taking advantage of.
    
    Re: 1.5 cents a minute, that's about a dollar an hour.  We can afford
    that :-)
                                                           

    Re: getting Plus! for Win95 to get dialup scripting - dialup scripting
    is also on the Win95 distribution CD under Admin.
    
    Regards,
    
    Garry

>    Re: 1.5 cents a minute, that's about a dollar an hour.  We can afford
>    that :-)

For 20000 employees who spend 100 hours a month on line it's $2,000,000
which would be free with high-speed modems.  Is ISDN speed really that much
of a productivity boost over 28.8 (and modems are getting faster)?

/john

    	Back to .0 get an ISP(independent service provider) typicaly around
    $20.00 a month unlimited access time at 28.8 bps. Next check out the
    http://tunnel.ibg.ljo.dec.com/tunnel-getting-started.html page on the
    internal web.

    It's not 100 hours a month, since this is evening and weekend
    use, normally, so we're ponying up a few dollars an evening in
    return for a few hours of an engineer's time.  Plus there are the
    previously mentioned volume discount plans.  Plus, in this area,
    there's a usage charge pending PUC approval for "non-ISDN" business
    lines of a similar amount.       
    
    ISDN speed is 128K, 4 (more or less) times 28.8K;  it takes
    an eternity for the graphically oriented apps we use to update over
    28.8K, to the point where they're basically unusable.  So for us,
    without ISDN, people normally wouldn't be using home connections for
    anything other than occasionally checking email.
    

    > For 20000 employees who spend 100 hours a month on line it's $2,000,000
    > which would be free with high-speed modems.  Is ISDN speed really that
    > much of a productivity boost over 28.8 (and modems are getting faster)?
    
    I'll bet most of the GMA employees wouldn't be interested in dialing
    from home at all, and of those remaining, few would spend 100 hrs/mo.
    For many, 28.8 is probably fine.  For me, running eXcursion with
    multiple emacs and xterm windows, it's very painful.  Painful enough
    that I don't do it much.
    
    Modems can't get much faster than 28.8 on POTS (Plain Old Telephone
    Service) lines.  28.8 squeezes out most of the available bandwidth. 
    Using FX lines to dial in, my 28.8 modem typically throttles down to
    about 24Kbps due to extra noise.
    
    There are other advantages to ISDN too, like high-speed dialing, and
    telco-guaranteed data rates (NYNEX makes *no* guarantees about what
    modem data rates you can get over a voice line, they just guarantee
    that you can communicate reasonably via voice).
    
    	Steve
    

>    For many, 28.8 is probably fine.  For me, running eXcursion with
>    multiple emacs and xterm windows, it's very painful.  Painful enough
>    that I don't do it much.

Why would you run an xterm window over eXcursion when you can use local
terminal windows (e.g. VTSTAR) for that sort of application?  I have five
of them up right now and they're just about as fast as when locally
connected.

>    Modems can't get much faster than 28.8 on POTS (Plain Old Telephone
>    Service) lines.  28.8 squeezes out most of the available bandwidth. 

That's what people said when 1200 first came out, yet modems are getting
faster and compression algorithms are getting better.

I don't believe that eXcursion at 56Kbps ISDN (no compression) is going to
perform much better than at 28.8 Kbps with compression.

100 hours a month is only a little over 3 hours a day.

/john

Re: .12

What is VTSTAR and how do you run it on the PC and the host?

Gim

    
    VTSTAR is an internally developed terminal emulator that was
    that was cancelled before becoming a product. The folks who 
    created it still support it (as time permits) for internal use.
    
    VTSTAR runs on the PC and connects to the host via tcpip, modem,
    lat or cterm.  Over tcpip (including across modems) you can run 
    multiple terminal windows... or you can get multiple terminal
    sessions over a serial line by connecting to a terminal server
    with TD/SMP support or a host with SSU (host based TD/SMP support).
    
    Available for Windows, WindowsNT and Win95. 
    
    See ONTIME::VTSTAR  and ONTIME::VTSTAR_BUGS.
    
    								- Mac
    
    
    	(This note written in session 4 of 6 using VTSTAR.)
                                     

John:

> >    Modems can't get much faster than 28.8 on POTS (Plain Old Telephone
> >    Service) lines.  28.8 squeezes out most of the available bandwidth. 
> 
> That's what people said when 1200 first came out, yet modems are getting
> faster and compression algorithms are getting better.

  The difference, of course, is that phone audio traffic is
  transmitted over digital links that are about 56Kb/s wide,
  and as long as that's true, then, measured in raw bit rate,
  *NO ENCODING OVER A VOICE LINE* will ever go faster than that.
  In fact, we can only asymptotically approach that rate, so
  28.8 and the newer 33.x modems are just about all that can
  be achieved on real phone lines.

  It was a lot less of a stretch from 1200 to 2400 to 9600.

                                   Atlant

> VTSTAR is an internally developed terminal emulator that was
> that was cancelled before becoming a product. The folks who 
> created it still support it (as time permits) for internal use.

  Actually, the one folk who created it (Jeff Lomicka) left for
  greener pastures a while ago. There may have been others who
  worked on the team as Whack was productized into VTSTAR, but
  the lion's share of the credit belongs to Jeff and Jeff alone.

  There is no official support for VTSTAR.

                                   Atlant

    
 > Actually, the one folk who created it (Jeff Lomicka) left for
 > greener pastures a while ago. There may have been others who
 > worked on the team as Whack was productized into VTSTAR, but
 > the lion's share of the credit belongs to Jeff and Jeff alone.
    
    No to get drawn into a spitting contest but...
    
    Actually, a great deal of Whack was re-written and a great deal
    of work was added by a team of (at any given time) about 8 engineers
    writing code over an active period of about a year. Having sat in 
    the engineering meetings with all of them day after day during the
    development of VTSTAR, I certainly acknowledge Jeff's accomplishments
    but "the lion's share of the credit belonging to Jeff and Jeff
    alone" IMHO does a disservice to the team.
    
    								- Mac     

    VTstar also runs on MAC and NT/Alpha. Writing this on VTstar/NT/x86.
    
    I believe voice data doesn't get sent at 56kbit for any distance these
    days. It gets data-compressed down to a lot less if you go over a
    trunk, eg. 8kbit. Given that, you need local access to the other end of
    your modem call or your 28k8 modem is useless (or do the exchanges
    recognise v.34 and pass it digitally - I doubt it).
    
    'Course, the whole thing may be academic. The isp I use rarely gets
    more than 4800baud actual thruput to my PC. 28k8 modems are wasted.
    
    regards
    john

    re .18: I don't think it usually gets compressed anywhere.
    
    

  Compression to about 32 Kb/s (about half the normal data rate)
  is becoming common here in the US. I *THINK* it's done on our
  very own DTN, but John Covert can give an authoritive answer.

  I'm not aware of anywhere that voice is routinely compressed
  down to 8Kb/s. Speech still gets pretty "vocoder"-sounding at
  such low data rates.
                                   Atlant

    Hmmm.. as all of Germany is ISDNized now, and ISDN guarantees me 64
    kbps, they couldn't compress that much further if I'm transferring
    already compressed data.
    
    As far as I know, the only difference between an ISDN connection and a
    POTS connection here is whether the codec (or whatever it's called in
    ISDN) is at my home or at phone company premises.
    

>  Compression to about 32 Kb/s (about half the normal data rate)
>  is becoming common here in the US.

On private networks only, where ISDN is never supported at all.

/john

    Also in regards to charges, here in the Chicago area, Ameritech offers
    ISDN service that is charged identically to POTS.  So if your local
    office is 8 miles or less than the local office of the Digital office
    you wish to connect to, you only pay a minimal call setup fee (~ 3 to 5
    cents) and no connect time charges.  I'd love to get ISDN service as my
    POTS lines are pretty poor at times.
    
    -tl
    
    

Anybody know why there's this 28.8K limit for modems, when specialised modems 
for analogue leased lines, such as BT's KiloStream, are quite happy running at 
64Kbits, and have been for years?  I have one sitting next to me, and the bit 
of wire it's connected to is just a normal telephone cable.

Chris.

>for analogue leased lines, such as BT's KiloStream

Kilostream uses a standard POTS cable to the nearest exchange (or some
point within 5km of you) where there is another "modem" that converts
to the digital hierarchy. Originally it used a pair for each direction
but the later models need only one pair (and were not initially as
reliable). (I think it's a baseband scrambler rather than a conventional
modem.)

You can do the same trick if you can (still) get a copper connected leased
line but otherwise any link will go through some form of multiplexing
equipment that limits the bandwidth and the quality. All modern switches
similarly limit bandwidth and quality. Over recent years as quality has
improved modems have been able to squeeze more bits/sec out of the 3.5kHz
(or so) bandwidth that is standard.

Peter

    re .24,.25: ISDN also uses a POTS cable from the central office to your
    home (or office) - and you get 128 kbps. (If you add the D-channel and
    the overhead, the actual transfer rate is even higher). It's what's
    beyond your copper wire that really matters (as .25 says).
    

    re .24
    Don't ask me to do the math anymore but to answer your question:  I
    believe that the chosen 'sampling' rate for vg (voice grade) lines is
    8KHz, if I remember correctly.  This places an upper limit restriction
    that a vg line can do to about 30-32Kb/sec.  Therefore, your 28.8 modems 
    are the best you can do and after that you're forced to another technology 
    to obtain higher speeds e.g., ISDN or ASDL.
    
    Can you get a higher rate from a vg line?  The answer is YES, however,
    to expect the telelphone companies to replace their existing equipment
    to sample at a higher rate would require billions of dollars of
    equipment world-wide.  Therefore, it more cost effective to introduce a
    new technology to run over the existing telephone equipment, with
    slight modifications at the central switch.
    
    There was an article regarding this in one of the Networking Magazines
    within the past 6 months.  Therefore, if you're interested in the
    particulars I suggest you might use AltaVista to research this further
    or look in the modems conference for something more concrete.
    
    	Regards,

Ta for the last few replies, I think I understand!

Chris.

    re .27: The standard digital transmission used in Europe (for ISDN and
    otherwise) i.e 8 kHz sampling frequency, 8 bit samples (64 kbps),
    happily carries 33.6 kbps analog modem signals (i.e. using an ISDN line
    and an a/b adapter at your home).
    
    Though I don't think you can do much better than that...
    

    Just happened across this article:
    
    The truth about high-speed modems
    January 1996 Windows NT Magazine pg. 40
    
    ...the fastest practical speed is about 28K bits per second (Kbps). 
    The 28Kbps is a function of the public switched telephone network's
    signal-to-noise (S/N) ratio.  (The laws of information theory -first
    proven mathematically by Claude Shannon in 1948 - determine how much
    information can pass through an analog channel with a given S/N ratio.)
    Most of the US has a S/N ratio of about 1000:1 for voice grade lines,
    which - according to Shannon's Law - yields a maximum data rate of
    about 30Kbps. ...
    
    Later the article mentions that with compression you can get effective
    rates of 38.4Kbps, 57.6Kbps, and even 112Kbps.
    
    Once again, to improve the S/N they would have to use a higher sampling
    rate which means changing Billions of dollars of equipment.
    
    	Regards,

    I was incorrect about the sampling rate:
    
    LAN Magazine July 1996 pg. 81
    Multimedia LANs and WANs
    
    ...  The Nyquist Theorem, a well-known communications theorem, states
    that in order for a sampled data set to unbiguously represent a
    continuously varying analog signal, the sampling rate must be at least
    twice the highest-frequency compoment in the analog signal.  Because
    telephone systems were designed to have a bandwidth a bit less than
    4KHz, most modern telephone system sample the analog signal at a rate
    of 8,000 samples per second.  Eight-bit converters are used, so the ADC
    outputs binary data at a rate of 64Kbs (8 multiplied by 8,000). 
    However, this is where the S/N value comes into play and limits this
    value to 1/2 the theoretical max or 64Kbs/2 or 32Kbs.
    
    I have yet to find the full formula.  One day I'll happen across it
    again.
    
    	Regards,

    8 bit samples give an S/N ratio of somewhere in the order of 45-48 dB
    - I can't say off the top of my head how many bps this would equal. But
    from experience, I know 28.8 kbps is no problem and according to some
    tests I've seen, you can achieve 33.6 kbps with good a/b converters via
    ISDN.

    We're not arguing that the typical phone line won't do 28.8K or 33.+K,
    that only an 8.75% difference and won't make much difference in your
    on-line activities.  However, if the regular phone system could handle
    56K or 64K (94.4% and 122.2% respectively) that would make a BIG
    difference in your on-line activities.  It probably could if we just
    lower our expectation e.g., S/N 500:1.
    
    	     Regards,

    re .33: Well I'm not really trying to argue anything. I find it rather
    amazing myself that a digitalized phone connection (like ISDN) is able
    to carry analog signals of these speeds.
    
    And regarding the regular phone system carrying 64k - depends on what
    you call regular; ISDN does it, and it can slowly be considered
    "regular" in many places in Europe. (In Germany, apart from the
    one-time connection fee, it costs exactly the same monthly charge as
    two POTS lines, so it really is the same price).
    
    Yes, I just ordered one a few days ago... while German Telekom is still
    subsidizing new ISDN connections with DM 700 (~US$ 450).

    re .-1
    I didn't mean argue in the negative connotation.  I should have said 
    discussing the features, benefits, and limitations of ISDN, etc.

    In the US they have waived the installation fee, which I believe is
    suppose to be about US$ 250.  The monthly charges are between US$
    30-40.  They made a mistake and didn't realize that people would get on
    the line and just stay there forever.  They thought ISDN would be used
    like a voice phone - few minute call etc.  Therefore, they're change
    the rates constantly.  I believe they're charging between US$ 0.015 to
    US$ 0.02 per hour between 8-5 and free anything around those hours.  I
    think they're trying to raise those rates about 35% (average) in the US.
    These numbers differ if the phone is registered as Business vs.
    Personal.
    
    Is Digital in the US going to start provding ISDN service instead of
    async modems?
    
    	Regards,

    E-1 is the standard in Europe, and T-1 is the standard in the US and
    some other places.  
    
    T-1's are time multipled 24 conversations.
    E-1's are time multipled 30 conversations.
    
    So assuming the standard frequency, I forget the actual the band width
    is greater in Europe, so the telephone should handle greater data
    rates.
    
    Dave

    re .35 Don't worry, no offence taken...
    
    Well over here in Germany (over there, to be exact, I'm in US just now
    :-) the rates are exactly the same for POTS and ISDN service (on a
    time/charge unit basis) so per bit ISDN is obviously cheaper. If you
    want to bundle both B-channels to get 128 kbps it's charged as two
    calls (as it would, obviously, if you use the two channels
    simultaneously for two voice calls).
    
    Another great advantage of ISDN is the quick connection establishment -
    no length modem handshake procedures (which also tend to fail more
    often than desirable). 
    
    re .36: I don't think that as any effect on the line speed - the
    channels are (always?) digitized at the rate of 8 kHz, 8 bits, as has
    been mentioned - no matter what's behind it.
    

>I believe they're charging between US$ 0.015 to US$ 0.02 per hour between
>8-5 and free anything around those hours.

In New England, a local ISDN call costs about that PER MINUTE regardless
of time of day.

/john