Tue, 24 Jan 1995 20:10:43 +0100
|
On Tue, 24 Jan 1995 13:54:56 EST Will Sadler
<[log in to unmask]> said:
>This discussion brings up something I have wondered about. Why does
>BITNET even still exist? I was under the impression that the
>optimizations that LISTSERV can do linking LISTSERV sites together can
>be done under BITNET or SMTP. Does LISTSERV require BITNET for this
>feature?
There are two flavours of LISTSERV: LISTSERV-NJE and LISTSERV-TCP/IP.
LISTSERV originated in 1986 as a VM application that required BITNET (at
that time the Internet as we know it did not exist, it was a limited
project that was mostly restricted to defense sites in the US and a few
cooperating universities). In March 94, a TCP/IP version was developed
and the old version was renamed to LISTSERV-NJE for clarity. In June 94,
this TCP/IP version was ported to VMS and unix. The TCP/IP version can in
fact run in NJE compatibility mode if desired, so if you buy LISTSERV
today you always get the TCP/IP version. But about 250 sites had gotten
the NJE version while it was freely available, so there are many active
copies of LISTSERV-NJE. Some are migrating to LISTSERV-TCP/IP on VM, many
are migrating from VM to unix or VMS. Most however do not see any need to
migrate until their organization's overall plans call for phasing out the
mainframe system or dropping the NJE connection. Some of the unix servers
for instance jumped from 0 to 220 lists overnight, as they reached the
corresponding item on their migration schedule/checklist. The migration
policies vary from one organization to another. Some want to make all
changes overnight, so that users only complain once :-), while others
prefer to migrate one service at a time, due to lack of manpower. It
seems most mainframes are going to be phased out this year, either this
summer or at Christmas. The ones that will survive into 96 are mostly
from places that have no intention of getting rid of their mainframes in
the near future. I think about 1/3 to 1/2 of the current VM sites are in
that situation.
The problem we are having with BITNET is that the core's capacity is
having trouble growing at the same rate as the traffic. CREN has a plan
to address that problem but before I comment on it I have asked them to
provide some sort of official description of the plan as I don't want to
misrepresent them or anyone else. Let's just say that I am fully aware of
the problems and that I am not sitting here waiting for a miracle to take
place. These delays have been mostly tolerable thus far but it is now
becoming apparent that we must find a solution that can be implemented
say within the next 3 months, as opposed to just letting the slower
migration to TCP/IP take care of it.
Eric
|
|
|