You can post content to the Google Search Appliance (similar to Google Site Maps for the GSA).
GSA Feeds Protocol allows you to add meta data to each record, which can then be used to filter search results.
You can feed the content of the document into the index as well, which allows you to control the actual content that’s indexed. Interestingly, you can post binary data formats as well.
You can group content into “collections” which are based on URL patterns.
XSL stylesheets for search results (already knew this. You can also query via the GSA’s XML interface).
Based on URL patterns you can mark some content secure, and others unsecure. Along with an access parameter to restrict search results. Additionally, the search appliance can ping a callback URL via HTTP GET with the user’s cookies and show results based on response (401 = unauthorized, 200 = authorized).
The GSA’s now allow you to install “OneBox”‘s (e.g. weather, FedEx, etc. data from Google) that shows dynamic information in search results. Triggered by a keyword and then grabs information from a provider URL based on that query. OneBox’s trigger an HTTP request to the provider URL, which returns XML that is then translated by your OneBox’s results XSL template. Insanely cool.
GSA’s can do up to 30,000,000 documents and beyond with custom setups.