Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!easy!lron
Newsgroups: comp.sys.amiga
Subject: Re: UUCP Summary
Message-ID: <186bb776.ARN08b4@easy.hiam>
Date: 26 Dec 90 00:01:42 GMT
References: <1869097d.ARN0cac@cbmami.UUCP> <37219@cup.portal.com>
  <186abac2.ARN08a8@easy.hiam> <37243@cup.portal.com>
Reply-To: uunet!easy!lron
Followup-To: comp.sys.amiga
Organization: You must be talking about someone else.
Lines: 77
Expires:
Keywords:
Distribution:

In article <37243@cup.portal.com>, Thad P Floryan writes:

> lron@easy.hiam (Dwight Hubbard) in <186abac2.ARN08a8@easy.hiam> writes:
>
>       [...]
>       No, way more like 700cps.
>       [...]
>       I'm connecting with a USR HST in V.32 mode with V.42bis and am seeing
>       about 650cps.
>
> Though I already answered this, there are two more items one may wish to
> consider:
>
> 1. I have a LOT of UUCP traffic in-out of my systems ("thadlabs") and I
> periodically enable full debug mode (via shell script (this IS a UNIX box)) to
> sample UUCP's song-and-dance.  I've often noted that many "far-end" V.32
> modems do NOT have MNP 5 and seem to have only MNP 3 or MNP 4.  Point being:
> though Dwight's modem is V.32/V.42bis, the "other" modems might not be, thus
> the "slower" transfers he experiences.  One of my connections is to a major

Both my modem and the Telebit I connect to have V.42bis and it is enabled during
connects  (link diagnostics on the  modem show  9600 connect  with LAPM/V.42bis)
Most people with V.32  modems I know disable MNP-5 because it increases transmit
time on compressed data.   V.42bis and MNP-7  and above don't have this problem.

> back-bone in Palo Alto (4 miles away) and it only has MNP 4 on a Codex modem;
> I get better through-put with sites on the East Coast and even outside the
> country at MNP 5, 7 or 9.

Yes, but your probably not using compressed batching either.  I got much
higher rates before we started using compressed batching but the overall
transfer times was longer.

> 2. The UUCP protocol used by AmigaUUCP is *NOT* the best: it is UUCP "g"
> protocol.  I now most-often use "e" protocol (when feasible).  I "discovered"
> "e" protocol two years ago when first setting up StarLAN (1BASE5 802.3e-1988
> over twisted-pair) in my lab.  I was extremely unhappy with the 2000 cps
> throughput using "standard" UUCP and a colleague recommended HDB UUCP with "e"
> protocol which I now use and enjoy an average of 50,000 cps over the same net
> with NO other changes (i.e. keeping HDB and going back to "g" protocol again
> nets only the 2000 cps).  The protocol is specified in the Systems file (or
> the L.sys file with the older UUCP versions) though it can also be specified
> in the Devices files.  Maybe someone has the time to implement other protocols
> in AmigaUUCP?  Note that one SHOULDN'T use "e" unless one has GOOD modems (or
> a good network) and can guarantee reliable end-end connectivity.

Tell me about it.  We are experiencing  satilite delays  around a second on my
only major long distance connect (Hawaii to Virgina).  At 9600 it doesn't take
the modem a second to transmit all 7 frames and the modem sits inactive as the
ACKs go back over the line.  This is the  same problem that was being discused
about a month ago on in  comp.dcom.modems.   I would love to get a hold of the
'e' protocol for  the Amiga,  but I understand you need an AT&T source license
to get the source for it :-(

> One additional note (though I posted a complete set of these specs several
> months ago) from the Microcom manuals (they invented MNP):
>
> MNP 5 Class 5 service provides data compression, which combined with Class
>       4 allows throughput of almost twice the connection speed.  MNP Class
>       5 dynamically adjusts to the type of data being transmitted for
>       maximum compression efficiency.
>
> MNP 7 Class 7 service provides Enhanced Data Compression, which combined
>       with Class 4 allows throughput of more than twice the connection
>       speed ...
>
> MNP 9 Class 9 service includes all the features of MNP CLass 7 Enhanced

MNP isn't bad (we have several modems I work with that have it).  But most
companies  don't license  any of the MNP levels above MNP-5  since MNP 1-4
are all that are needed for V.42 compliance.

-----------------------------------------------------------------------------
| Dwight Hubbard,                                       |-Kaneohe, Hawaii   |
| USENET: lron@easy.hiam or uunet!easy!lron             |-GT-Power: 029/004 |
-----------------------------------------------------------------------------