web cache and remote access

[Karen G. Schneider]

"This weekend, they implemented something to the effect of a "web caching
server" without any notification. As a result we immediately lost access to
all of our subscription online resources that authenticate by IP, because
the server altered the IP number..."

I'm coming late to this thread, but I don't see this as a technical
issue--it's a human issue.  The problem seems to be less about caching,
proxying, etc. than about coordination and communication.   Not only did
they not NOTIFY you, but they didn't find out what your needs were ahead of
time.  It was their job to find out where the exceptions and problems would
be, and work through these issues before implementing this tool.  The
exception list for the databases should have been in place before changes
were made, period. The bottom line is public service.  (To contrast: our
consortium is implementing some changes that have similar impact, and the
systems people had a meeting, has done site visits, consulted on the
potential problem areas., etc.  Subscriptions are one of several key areas
they are concerned about.  Inevitably, there will be problems, but they have
done Herculean efforts to keep them to a minimum.)

Another comment caught my eye: "It seems to me that, for most libraries, the
opposite is true - 80% or more
of a libraries' Internet activity goes to online subscription databases..."

This is where analytical tools come in handy.  We get statistics from our
online vendors, and we also use Webtrends to analyse internal and external
vistors to our websites.  Overwhelmingly, the top resources are general
Internet search tools such as Yahoo, job resources, and email websites.  Our
online databases are used, but not heavily, even though we stack the decks
in their favor by having three machines dedicated to their use in addition
to providing these databases on our fourteen general-Internet PCs.  (They
also aren't accessed that frequently from home, either.) Maybe things are
different in an academic library, but I know of at least one very large
research library that has similar statistics.  For that matter, though we
offer word processing on ten of our PCs, according to the statistics
generated by sign-up sheets, Internet is preferred to word processing 5 to
1.  (Though folks using word processing stay on a lot longer too, so that
might be a confounding factor.)  So you might want to use a tool--Webtrends,
Site Server Express, etc.--that measures what's really going on.

Karen G. Schneider kgs at bluehighways.com
Assistant Director of Technology
Shenendehowa Public Library, Clifton Park, NY
http://www.shenpublib.org