At 03:33 PM 6/10/2004, Stephanie Kripps wrote:

Hi everyone:  I am collecting info on the experiences of large or medium
*public* library systems that are using any oss. (In North America -- not
NZ, Europe or elsewhere). I specifically want to know what obstacles and
glitches had to be overcome.
I already have seen some material written by Cindy Murdock from Meadville
Pub Lib, as well as material about Howard County Pub Lib.  Are there any
other large-ish public lilbraries out there with an experience?
This will help our own decision-making process.  We already know about the
benefits of oss -- but what surprises might we encounter?  What to beware
of?  from the public lirbary perspective.

We first started using OSS when we began building out an Internet presence from our previously private library consortium WAN in 1995. After much research and testing, FreeBSD was selected, first to build a firewall for our (wow!) 128K ISDN Internet feed, and later to build our own internal DNS, WAIS and email servers.

The WAIS server barely lasted beyond pre-production testing, simply because GUI HTTP browsers came along and stomped a mudhole in the popularity of older protocols. A bit later on, the role of web server went to an HP-UX box, due to ILS vendor integration requirements (which the vendor promptly reneged on by dropping their support of HP-UX as a platform for their web catalog; buh-bye $$$...). At least it ran Apache pretty well until HP dropped hardware support for that server model. We currently have multiple web server platforms using Apache on FreeBSD.

That first firewall lasted through a transition to T-1 line speed and later an upgrade to dual T-1 lines, and though the box has been wholly replaced once, and partly replaced twice, the model has maintained the sacred "five nines" of uptime for seven of the last nine years. Our first email server ran with nearly five nines of uptime on a refurbished 486 with 80 MB of RAM, from 1996 to 2003.

Surprises? I continue to be surprised at how much better prestige is than profit, as a driver of quality. But you're asking about the nasty sorts of surprises no one wants to have. I usually find that having proper expectations helps keep nasty surprises to a minimum, so:

Don't expect to adopt brand new hardware models before checking with your OSS community for the availability and robustness of drivers. New chipsets, new motherboards, new graphics cards, etc. can give early adopters a bit of a headache. Let others be on the bleeding edge.

Don't expect that OSS will necessarily make life easier. More affordable, stable, secure... quite probably. However, in many cases you'll be called on to roll up your sleeves and get your hands a little dirtier poking under the hood of your systems. Some of the end user and administration interfaces will seem a bit raw, and you'll find that OSS documentation isn't always as polished as that for commercial products, but on the other hand it's often more accurate, detailed and honest than its commercial counterparts.

Don't expect that commercial library market vendors will fully understand your desire to be freed of certain product brand name and version number requirements. It's getting better but it seems like a long slow dance of three steps forward, two steps back. For this reason, we are still primarily a Windows shop as far as end users are concerned. I think Linux/BSD on the desktop is probably at least 3-5 years away for us, and certainly the evolution of DRM is going to put quite a spin on the subject of what is needed and desirable on the desktop in coming years. This is one area where libraries of all kinds may be in for some nasty surprises.


--------------------------------------
Greg Barniskis <gregb@scls.lib.wi.us>
Library Interchange Network (LINK)
South Central Library System (SCLS)
(608) 266-6348 www.scls.lib.wi.us