In fact all the big shops use Solr searching and not Drupal's built in search. Awesome no?
Including, in fact, the White House, which is on a LAMP stack of Open Source goodness, including Drupal and Solr. Awesome indeed.
For predictive search, you'll want to get friendly with the Solr TermsComponent, which serves up the terms present in your index along with their frequency.
If you want to get really fancy, you can log your popular queries—particularly the ones that have a high correlation with click-throughs.
Works fine, but hard to keep data updated.
That depends on how you are interacting with Solr. There are a number of good clients that integrate in to popular ORMs and can automatically post over updates to Solr as data changes in your application. I'm compiling a list of popular Solr clients over at https://websolr.com/guides/solr/clients.
For some popular example, there is RSolr or Sunspot for Ruby applications. Haystack is a good one for Django, and there are Drupal and Django extensions as well.
"A child is a person who can't understand why someone would give away a perfectly good kitten." -- Doug Larson
