Fermilab's First Public Webserver Retired...

Technocrat

Fermilab's First Public Webserver Retired...

mengel

Wed Nov 26 11:39:00 -0800 2008

Computers.

Last editor rubycodez

manage

For those of you who care about such things, www0.fnal.gov, the Sun Sparcstation 10 desktop box which was Fermilab's first public webserver, set up in 1994, was retired.

It originally had a page that looked like: FERMILAB COMPUTING DIVISION...
in the original text-mode "www" browser.

It had an uptime of 327 days at its last shutdown.

Edit by Rubycodez: The webmaster of Fermilab's first public webserver, Dr. David Ritchie, helped to clear up a little confusion. The webserver of this article, a Sun box, was the first public one at Fermi National Accelerator Laboratory. But the first web server at Fermilab, set up in 1992, was a Digital Equipment Corporation Vax.

On Sun, Nov 30, 2008 at 10:59 AM, David J Ritchie <xxxxx@xxxx.xxx> wrote:

Dear Ralph,

According to a document which I co-authored with Paula Garrett in 1995... "Collaborating over the Web: Libraries and laboratories or the Librarian and the Webmaster", by P. Garrett and D. Ritchie, March 1995, http://lss.fnal.gov/archive/1995/conf/Conf-95-056.pdf,...

"a server to provide online data acquistion system documents was set up by Tim Berners-Lee, CERN, and Jonathan Streets, Fermilab, on a visit by Berners-Lee to Fermilab in July 1992.

That first server ran on a VAX/VMS system and used DCL (Digital Command Language) in its implementation...This 1992 technology would be considered fairly primitive by today's standards. [March 1995]"

The above link is in an authoritative library collection but it points to a scanned image of the document and the title isn't very helpful so it is easy to miss when searching for history of the web.

Here are some useful links about web history...

Here's a full text version which I think is close to the scanned version which I will ask our library to catalog: http://home.fnal.gov/~ritchie/ANCIENTtasks/presentations/ausweb_95/AW02-03.html

A graphical time line of Fermilab: http://cdorg.fnal.gov/com/spiral.html

Bringing the Web to Fermilab: http://home.fnal.gov/~ritchie/OTHERtasks/Bringing%20the%20Web%20to%20Fermilab_files/frame.htm

I do remember a Sun being used during an "Industrial Affiliates" presentation in 1994 or so. Also, the CHEP94 conference had some use of Sun's...

--
David J. Ritchie

ultra 10????

rubycodez

Wed Nov 26 14:09:52 -0800 2008
manage

Th e Sun Ultra 10 was introduced 1998, so no way was it the model used in 1992

citation for story, maybe a misprint?

ultra 10????

qcubed

Wed Nov 26 14:35:53 -0800 2008
manage

It's at Fermilab; apparently the particle accelerator punched a hole through space-time enabling this box to come through a few years early.

ultra 10????

rubycodez

Wed Nov 26 14:48:06 -0800 2008
manage

heh, well I'm e-mailing Fermilab's first webmaster for the skinny on what machines were first used for the 1992 website and then the 1994 public one, I can't remember. Likely was a vaxstation, we'll see.

Edit: First one was a Vax! I've edited article, that was first public webserver that was taken down, Sun Sparcstation 10.

ultra 10????

Michael Wagner

Wed Nov 26 14:49:24 -0800 2008
manage

Maybe they confused it with the Sparcstation 10?

That one came out in spring 1992...

ultra 10????

rubycodez

Wed Nov 26 15:13:03 -0800 2008
manage

that could be, maybe Mengel might have made a little slip in story submission. heh, I just realized I think I know who he is if it happens he actually works at Fermi. 8D

ultra 10????

rubycodez

Sun Nov 30 09:22:34 -0800 2008
manage

you're right, it was Sparcstation 10. Also, it was first *public* webserver at Fermilab, not the first webserver at Fermilab which was a Vax.

Third ever webserver retired...

mcrbids

Wed Nov 26 22:31:04 -0800 2008
manage

I'm surprised that Sun hasn't purchased it and put it back online as an example of how reliable their equipment is!

Honestly, just combining Linux and best practices in hardware has enabled me to sustain excellent uptimes approaching 4 nines - when I quote my uptime stats for the past four years (about 99.96% - between 3 and 4 hours of downtime per year) I routinely get grilled about what that "really means".

When I explain that the numbers are just what they seem - an average of less than 4 hours per year of unexpected downtime per year 24x7, with an allowance for software upgrades that don't happen before 7 PM PST, the response I usually get is incredulity.

Apparently, numbers like this are unusual. I dunno - I delivered numbers like this for years as an ISP sysadmin armed with craptastic consumer-grade hardware, a few 5 inch "emergency" cooling fans, and a few Linux ISOs burned to CD...

It's not hard if you just don't dicker needlessly with your servers once they're running! (as in: DON'T FRICKEN' TOUCH IT!)

Third ever webserver retired...

Michael Smith

Thu Nov 27 01:24:49 -0800 2008
manage

Internet style applications are well debugged, possibly because the people who work on the operating systems seem to have a hand in application development. Consider the LAMP stack where OS and apps are considered equal partners in the whole system.

Its a different story with less well behaved applications. I agree with your point about leaving well enough alone. I usually install a current stable release of NetBSD, boot it up and uptime is determined by the mains power supply.

The only real problems I have are with dodgy drivers. Wifi drivers and their associated hardware have caused a few kernel panics for me. Stock ethernet interfaces are rock solid.

Third ever webserver retired...

Calum

Thu Nov 27 07:09:46 -0800 2008
manage

"It's not hard if you just don't dicker needlessly with your servers once they're running! (as in: DON'T FRICKEN' TOUCH IT!)"

Yeah, cos the security updates just apply themselves.