[boulder.pm] mod_perl

[Walter Pienciak]

On Thu, 17 May 2001, rise wrote:

> On Thu, 17 May 2001, Walter Pienciak wrote:
>
> > The LAN/MAN standards release went off without a hitch. My
> > apache/mod_perl server never broke a sweat (I had a little access
> > handler installed that returned a 503 message to robotic agents not
> > known to me as existing customer setups),
>
> Were spiders/download agents a big chunck of the incoming connections?

Our logs show that large portions of our sites are mirrored unofficially
in a lot of places.  With the expected demand, I wanted to make sure that
people who wanted particular documents took precedence over people who
were just sucking down a library for reference.  I haven't analyzed the
logs very carefully yet, but I handed out about 3k "503" responses in the
first two days.  With a saturated network, I'm sure it made a difference.
Plus, the mod_perl thing was just so darned easy to write/install . . .

> > but the company's 2 network connections were saturated for most of 2
> > days.
>
> Congrats.  Would you say IEEE is taking this level of interest as
> vindication of the concept?

Yup.  I gave the network folks a month's heads-up.  One person shot
down all plans for ramping up capacity because "it wasn't going to
generate that much traffic."  Based on the gleam in another staffer's
eye as he related that story, I rather suspect the clueless one has some
unpleasant discussions ahead of him.

An anecdote:  Some of the network guys wanted me to throttle the
server back (I had it up to 110 active httpd processes at one point,
with keep-alives enabled, pumping out multimegabyte PDF files), but I was
cranky.  I thought "it was the most important thing happening on the IEEE
network that day."  I also noted that they ought to be happy, since it was
the first real test they'd had of the redundant network connections.  ;^)
But they didn't say thank you, alas.

As far as other areas go, Theo DeRaadt wants POSIX (1003.1).  I tend
to agree it's a good choice.  But those two opinions == 0.   The reality
is that funding is the driving force (IEEE Standards are a self-funding
entity within the IEEE).  The Open Group takes its money up front in
large chunks, from corporations.  The IETF and the IEEE are open processes,
and the IEEE recovers its costs from the sale of the documents rather
than from grants.  The 802 pilot is an exploration toward the latter.

> > It was fun; now, back to the real world and less interesting projects.
>
> Clearly it's time for another hike.  Anyone interested?

Well, you know I am.  But you've gotta be getting bored with my ugly face.
Anyone else?

Walter