On Wed, 16 Sep 1992 01:54:44 -0400 "Terry Kennedy, Operations Mgr"
<TERRY@SPCVXA> said:
> While I've never tried moving a file between two European hosts, I can
>say that performance has been very good between those hosts and my
>system here in the US.
European hosts, and in particular Germany, have much better connectivity
to the US than with each other, because politicians generally accept that
they can put no pressure on the US regarding topology, protocols et al,
whereas anything within Europe is a hot political topic nowadays. I
consider myself lucky if PING to Germany comes back within less than
2000ms. BITNET files don't move faster, but they do move without my
intervention. Poland is connected to EARN at 9.6k (half-duplex) and to
the Internet at 40k (the plan is to run VMNET, but they are having
trouble getting a suitable ethernet controller). The Poles use the
network to order PC packages, PC packages and PC packages; occasionally
they may also send mail. In spite of the fact that the IP connection is
about 5 times faster, they are using BITNET to get their PC packages;
that line works 24h at 100%, whereas the IP one is about 40% used. Can
anyone tell me why one might want to use inferior protocols running at
1/5th the speed?
>> FTP can't even transfer a VMS saveset unless both systems run
>> Multinet. SEND/FILE/VMSDUMP works.
>
> I beg to differ. The homebrew FTP client and server I run on my VAX
>handle this format quite well
I stand corrected, it only works if the server is running TGV's product
and the client is either TGV's or a homebrew or adapted version. The
situation remains the same: concretely, when I FTP a package for VMS,
half of the time I have to get a separate "fixup" program which has to be
compiled and run manually on the file so I can use it at all.
>The maximum BITNET file size is 300K,
Oh, please! We all know that this limit exists only in paper, and that
the only reason it is still present (in the US) is bureaucracy. I send
and receive larger files every day without problem.
>I also get my BITNET files delivered to me at the lightning fast speed
>of over nine thousand bits per second!
Don't tell me you have never heard of VMNET and JNET's TCPNJE? What would
you say if I told you that I am planning a move from VMS to unix because
vi is more user-friendly than TECO? :-)
>Just last week I retrieved a version of an RFC-compliant SEND program.
Yes, and there are many problems with this protocol and its current
implementations. In particular, it will never work on machines with large
amounts of users as it is now. This may be a small amount of hosts, but
it certainly is a significant amount of users.
On the 8th of July, I informed the author of the VM version that I needed
two changes to the VM implementation in order to be able to support this
protocol with LISTSERV: the ability to request the server to cleanly
reject messages for which reverse lookup failed (alternatively if this is
not possible LISTSERV could toss them out itself, but it is more
user-friendly if the server rejects them as per the protocol) and a
non-blocking interface. This is a *minimal* requirement, I completely
ignored a number of other concerns I had with the protocol. It has not
been possible to implement these changes in time for 1.7d.
>While hosts that support it are scarce at the moment, all of our (SPC)
>hosts will, and I expect it to catch on. Likewise for sender-controlled
>file transfer.
You are being very optimistic. I doubt either protocol will really catch
up before 2 years (the latter is not even a protocol yet).
>> I am on BITNET and I can TELNET around. I must have missed something.
>
> But that's an Internet service. Having BITNET doesn't affect it at
> all.
Obviously. The person I was replying to seemed to say that one should
leave BITNET because TELNET is better than BITNET services. I just
pointed out the flaw in this argument.
Eric
|