That’s all there is to it. It’s the reason that in 2009 Google pulled down $17,863,013.70 in pure profit a day (wikipedia).  We’ve (We means the University) been using a Google Mini appliance for just about 6 years (It’s been here longer than I have) and it desperately needs to be replaced.  The problem is Google Mini’s are expensive, and not sustainable. Drop $9,000 today and in two years your support will run out, but don’t worry you can just buy a new one!

We decided to look into Nutch an open source web-search software. It works a lot like a Google MIni, you buy a piece of hardware and once setup properly Nutch does the crawling and  indexing.  Nutch showed a lot of promise and i was really excited to be out from under the Google Mini. The issue we ran into with Nutch is that search is hard and isn’t for hardware that is, well puny.  Nutch runs best on a Hadoop Cluster and this is noted clearly in the About documentation.  In our environment getting anything better than a Dell R210 is near impossible so even getting one box good enough to power Nutch just wasn’t feasible.

We finally started looking at Google Custom Search. We talked to some folks from Virginia Tech, who had a lot of good things to say about their experience over the last couple of years. We’ve implemented it, and much like they did we’re pulling in search results from web pages, people and places. The Google Custom Search is free, unlike Google Site Search which is more customizable and because we are a university we can legitimately turn off the ads.

I’m pretty happy with the new setup, especially the price and the fact that we don’t have to support another piece of hardware.