Introduction to FAST
FAST, aka Fast Search & Transfer (www.fastsearch.com) is a pretty interesting and important company. They have 3500 enterprise customers, a rapidly growing $100 million revenue run rate, and a quarter billion dollars in the bank. Their core business is of course enterprise search, where they boast great scalability, based on a Google-like grid architecture, which they fondly think is actually more efficient than Google’s. Beyond that, they’ve verticalized search, exploiting the modularity of their product line to better serve a variety of niche markets. And they’re active in elementary fact/entity extraction as well. Oh yes – they also have forms of guided navigation, taxonomy-awareness, and probably everything else one might think of as a checkmark item for a search or search-like product.
FAST is actually a Norwegian company, in the tradition of strong European linguistics technology. But due in part to a series of acquisitions, it has operations all over the world, including in Needham, MA. FAST originally came to prominence for the web search engine alltheweb.com, but eventually sold that off to Overture, which in turn was bought by Yahoo. (This is the claimed source of their scalability and grid architecture.) I gather there’s some sort of a non-compete agreement keeping them from getting back into web search, but they actually do index the web anyway, as a service for their enterprise customers – i.e., as sort of a preprocessor for the customers’ own web spidering and indexing.
I haven’t totally figured out that modularity, aka “pipelining”, and the associated verticalization yet, but here’s the general idea. There are actually three “pipelines,” for indexing, querying, and result set manipulations. Apparently, this goes well beyond the usual tokenization/ part-of-speech/part-of-document/etc. pipeline any serious search engine has – and I think also past the UIMA model — to the point that there are open APIs allowing the addition of more or less arbitrary logic. The easiest examples to think of would be additional filters, such as security filters applied to manipulate result sets.
A vertical solution, then, would include a choice of modules; tuning of the modules selected; likely some truly niche-specific code; and also some packaged professional services. Not surprisingly, these typically evolved from specific professional services engagements. (Nothing new there, in the text indexing industry or otherwise; indeed, it’s also true of large portions of Oracle’s product line.) Examples of niches include the usual suspects; for details see the company web site. Actually, there seems to be a further subtlety, in that they sliced the salami very thinly into a bunch of “solutions”, then more recently repackaged groups of solutions around fewer, larger “themes.”
More details later.
Comments
One Response to “Introduction to FAST”
Leave a Reply
[…] I had a call with Business Objects, mainly about their overall EIM/ETL product line (Enterprise Information Management, a superset of Extract/Transform/Load). But I took the opportunity to ask about their deal with Attensity. (Attensity themselves posted more about the relationship, including some detailed links, here.) It actually sounds pretty real. They also mentioned that there seem to be a bunch of startups proposing search as a substitute for data warehousing, much as FAST sometimes likes to. […]