I hate to boast but then you asked for it. Here are three stories:
1. School of Mathematics, TIFR.
In Brief: At the School of Maths, TIFR, Linux had an abortive start
with 0.11 but really established itself at version 0.99.10 with SLIP
connectivity. Version 1.0 and motive.math.tifr.res.in was "properly"
on the internet with full ethernet connectivity. Now there is a
network of Linux machines with mail, DNS, http, and other internet
servers among them.
People Involved: Kirti Joshi (kirti at math.tifr.res.in), Pablo Ares
Gastesi (pablo at math.tifr.res.in), Krish Patnaik (patnaik at tifrvax.tifr.res.in),
Sunil Mukhi (mukhi at ganesh.theory.tifr.res.in),
Kapil Paranjape (kapil at imsc.ernet.in).
2. TIFR Bangalore Centre.
In Brief: In 1993, this institute was connected to the ERNET via dial-up
UUCP through some brand of Unix (which crashed and hung frequently)
running on an x86 machine. Linux was installed on a newer machine and
a leased line to ECE Department at IISc, Bangalore put this machine
(ns.tifrbng.res.in) on the internet over CSLIP and 19200. There is now
a network of Linux machines (and Mac's) that are workstations for the
folks at this centre.
People involved: Jayant Udgaonkar (jayant at ncbs.tifrbng.res.in),
Mythili Ramaswamy (mythili at math.tifrbng.res.in), Vasudeva Murthy
(vasu at math.tifrbng.res.in), Upinder Bhalla (upi at ncbs.tifrbng.res.in),
Avinash (avinash at ncbs.tifrbng.res.in), Gopi Garge
(gopi at ece.iisc.ernet.in), Kapil Paranjape (kapil at imsc.ernet.in).
3. Institute of Mathematical Sciences, Chennai.
In Brief: In 1996, this Institute had a full network of Solaris and
Irix machines which was connected to the Internet via a VSAT link to
ERNET. Linux had been tried (in a manner of speaking) and left in
unimportant places (too buggy, slow,...!). Linux made a real entry
with Debian 1.2 when a mirror site (xxx.imsc.ernet.in) for the LANL
e-print archive was set up. Then came a squid-based proxy server
(pp2.imsc.ernet.in) and finally, when X-terms had to be replaced they
were replaced by Linux (Debian 1.3) Workstations.
People Involved: T. Jayaraman (jayaram at imsc.ernet.in), G. Subramoniam
(gsmoni at imsc.ernet.in), Madhavan Mukund (madhavan at imsc.ernet.in),
Rahul Basu (rahul at imsc.ernet.in), Kapil Paranjape (kapil at imsc.ernet.in).
The School of Mathematics Experience (a longer version):
Prehistory: In October 1991 this department had a i386/387 with 4MB RAM
bought ostensibly to help people "learn to program". We ran Minix and
Minix-386 (with Bruce Evan's bcc) for a while and fending off
complaints from the MS-DOS folks at the same time. Got uucp working
with a dial-up to the TIFRVAX. We then learned via the Minix mailing
list about a guy called Linus Torvalds (!) who announced that he had a
Kernel that booted---but we had Minix-386 so why should we bother!?
Our main headache with Minix-386 was that gcc was difficult to port
(though it *had* been done) and so many GNU utilities wouldn't
work. So we tried out Linux 0.11 and it bombed because we didn't
really understand the VGA choice menu (the 386 had an EGA and none of
us had fiddled with graphics before). Thus, back to Minix-386; till
along comes TIFR's Internet connectivity (1992) and the possiblity of
getting onto the internet with our box. So while other departments
with huge budgets were buying SGI/Apollo/Sun servers and workstations
we experimented with Fred van Kampen's TNET for Minix. At this point
Dinesh Sharma (of IIT/Powai nowadays) said he had this GNU-OS up and
running with networking and the works! I had heard of the GNU/Hurd but
was it already running---no such luck---better luck---it was Linux.
Early Linux: Version 0.99.10 was (if my memory serves me right) the
first linux version we ran; it was as neat as we imagined (remember,
the our only other OS installation experience was Minix-386 and DOS).
The main thing was it ran gcc and the first thing we did was to get
the GNU utilities (file-,shar-,gcc- excepts emacs we didn't have
enough hard disk space for emacs) and compile them all (it took a
couple of days) and we were proud to be able to install them in
/usr/bin instead of /usr/local/bin. We now had an internet ready
machine but no ethernet connectivity to the TIFR network! So we used
the SLIP stuff (Fred van Kampen again I think!) and ironed out a few
minor bugs---motive.math.tifr.res.in was born. There was a continous
fight to keep it alive because:
(a) People wanted to use the dial-in to the TIFRVAX to check
their e-mail (no use telling them we had a much better machine right
under their nose which could also receive their e-mail).
(b) There were always these DOS folks who would just switch
off and on the machine in the hope of getting a more familiar prompt!
(I should mention that these were the only time the machine actually
went down---we *never* had a crash---not even a network hang---after
the initial installation was over).
Mature Linux: Around the time Linux 1.0 was released we got regular
ethernet connectivity which allowed us to free the serial line. We
could then also get a feel of the rapid Internet-based speed of Linux
changes. We heard of folks running X-windows and such exotic stuff and
felt cheated that there was no X-server for EGA. But wait, you say
what about Xega (still exists on sunsite!)? Well, it never gave us a
stable screen. So we worked on getting svgalib working for EGA (done
on motive.math.tifr.res.in) and finally (with some work done on
ganesh.theory.tifr.res.in the first Linux machine in the Theoretical
Physics group run by Sunil Mukhi) we also got MGR working. So now we
could beat the DOS-ers on their own turf. We had a fulling functional
TeX/LaTeX installation with an editor, viewer and printer driver. The
dvips we had under DOS was hacked and no longer supported (esp. since
the afore-mentioned hackers were now re-grouped under the Linux
banner). The complete-ness of the TeX installation was our Unique
Selling Point to argue for a couple of new machines to add to our
Linux "network". The main head-ached was the editor---we never really
managed to convince folks to switch over to Vi, Emacs or Pico or any
of the many editors that come with Linux (Slackware in those days).
Linux Takes Over: Soon we had a DNS server for math.tifr.res.in and a
small network of machines (motive, moduli, lattice) which were running
NIS (server and client) had user accounts for e-mail, TeX (editing,
viewing, printing) and internet facilities. Of late a Dec Alpha and a
Dual CPU Pentium have come in---the DEC still runs Digital Unix but
one isn't sure for how long... Oh, yes! Motive still exists but it has
been upgraded of course; the body has changed but the soul (Linux)
keeps marching on...