Good luck getting a Google Search Appliance to parse MARC records and extract meaningful data.

What kind of searching are you trying to get it to consolidate? How do you expect it to score relevancy? Do you expect it to put item records from your library right up alongside web resources? How do you determine which book is more relevant?

The Google appliance does a good job at providing an easy-to-customize (via XSLT tinkering) search box that will catalog all your stuff. I know on the mini it wasn’t even possible to have it search anything that was protected with anything more than HTTP auth in the header – but I know the search appliance has more features for that kind of stuff. Don’t expect that part to be easy though.

The biggest downfall I see in the Search appliane is that the relevancy score is weighted so heavily on incoming links. When most of your pages at the bottom level have the same number of incoming links – it will fall back on metadata and page analysis.

It doesn’t have a magic wand to search your content for you. There is no such thing as a “magic relevancy” score that can compare results from a wide variety of sources. Somehow, one of those sources is likely going to be looked at as “more relevant” by the search engine just based on how the content is presented to the bot than another source, and that source’s information will come up much higher.

I finished up at BYU last year – but unless things have changed you still have a large and powerful Computer Science department that is full of talented students. Most of them are probably willing to work for pennies. Have them build you a Lucene-based search box that will do your dirty work for you. Heck – you could probably even get a professor to get kids to work on it for a class or something.

Buying a Google appliance is a good decision for someone for whom it is cheaper than developing one internally – or if you really want a cool-looking server to put in your datacenter. I don’t know if the Google appliances come wiht T-shirts, but the mini sure does

I guess the other question I have is – who provides your library software? If they can’t deliver a search interface that handles all your assets, why are you using them?

http://swish-e.org/
http://www.htdig.org/