So they are thinking of implementing Sphinx at my work for large scale full text searching. Has anyone used this and if so how does it perform (against the DB its querying, not its own box its running on)? Looks really cool.

Yeah it looks interesting. I recently spent time looking into a search solution for the project I'm working on. I must look at Sphinx in more detail now.

I mainly looked at ElasticSearch and Apache Solr. They're both based on Lucene. Both have RESTful APIs and lots of native implementations. The both offer all core search bells and whistles.

ElasticSearch is designed to be a highly configurable distributed network with lots of replication/duplication. The documentation isn't the greatest though.

Solr doesn't scale as well. It's got a bigger community and better documentation.

From an initial look Sphinx seems to have a good mix of their best qualities.

I installed it on my vbulletin site and have been pleased.
Very fast and provides relevant results.
Full index of 4 million posts took about 10 minutes. Re-indexing with the vbulletin built in search would crash after about 4-5 hours, which is why I was looking for something else.
I got the sphinx server daemon up and running in about 5 minutes on my freeBSD box.
Server is an old dell 1950 with dual quad core 3.0 xeons.