We are all in the machine room:

	marthag ghudson rjbarbal autumn mhbraun mkgray mhpower tlyu |
jweiss yandros abbe shabby 


Matt opens.  Apparently greg did something right in the last 24 hours.
Greg disagrees.  We talk about proposals. Jeff likes short proposals
with lots of graphs, not more often than once every few months or so.

Matt suggests we collect data on servers.  We've been doing accounting
on news for a while. And on WWW.  Are we doing anything useful?  Eh...
not really.  We can make up pretty graphs. What would be appropriate
"metrics" (to use tqm rhetoric) for us to measure?

	number of connections
	max number of simultaneous connections
	throughput
	number of {articles, transactions} requested

Greg points out that we don't have anything close to exponential
growth. mkgray offers to write a perl script. (to crunch data, that
is. =)

mhpower says that all of this will not benefit us unless IS thinks the
services are important.

people spend a lot of time justifying web service. The basic point is
that if we can show that people want these services, IS will want them
to continue.

There is some discussion about censorship. Basic point: it is to IS's
advantage to have SIPB, a volunteer group, maintain news.

More talk. Jeff wants a contact for the services we provide.  Issue:
Jeff doesn't seem willing to talk to anyone but tytso, warlord, and
mhbraun. We think Jeff needs to delegate; he thinks we should have
someone personally accountable. mhpower argues that we don't want to
implement Jeff's idea of personal accountability.  There should be a
difference between a contact and a scapegoat.  We are a volunteer
organization; we can't be around 24 hours to respond to pagers.  We
make sure that things stay up.  How can we communicate this attitude
to IS?

What we need to do:

	Come up with a plan for dealing with crises
	Come up with a way to present our way of maintaining services
		to Jeff

Moving on....

Out of whose budget does our hardware come?  Kim's, and thus
indirectly through Jeff.  Ultimately, all our budget comes from the
academic computing pool (we think).  Kim might be willing to be the
contact for sipb, considering most of our recent hardware has come
from her pool.  Lots of positive feelings about this. It's good
politically and non-politically. How about Greg Jackson?  mhpower
argues the opposite side: Jeff has been around a long time and will be
around for a while. Also, Jeff is a sipb member and has supported us
for a number of years.

Abbe points out that we are not helping the problem any by excluding
people who might possibly be interested.  Should we have announced
this meeting to the sipb as a whole?  People point out that they had
no intention of excluding people, they simply want to reduce the
flameage level.  Abbe's point should be discussed further, at the
meeting.

mhbraun points out that one thing that we should consider is the
reliabilty of our data.  This is why charon's discuss spools are
backed up to AFS land nightly (picayune is not because of lack of
space).  This puts charon discuss data reliability on par with
Menelaus's 

SIPB tends to be historically bad at predicting what hardware we need
and when we need it.  This needs to improve (possibly with better use
of long term data) 

IS is not taking us seriously now in some ways....some say this is
becuase we are falling down.  ghudson points out the reason things are
falling down is that they (IS)  are not being responsive to our needs.

IS has never refused our machine estimates and we have always been
overly conservative in our estimates.  We should think more carefully
about our growth so that we can estimate better.  

We need a server hardware proposal that will satisfy our need for now
and the future.  (yandros will deal) 

Matthew and someone should go see Kathy and discuss these issues.
THis should be announced at the meeting. (and be political) 

Machine room stats: Matthew (and whoever volunteers to help) will deal