>I'm hoping sooner or later database companies realize that httpd servers are
>really just internet front-ends to public or private databases and start
>bundling or selling httpd servers that sit directly on top of their
>technology - not cgi-bin gateways, but an integrated part of the database
>itself. Anyone doing something like this for Sybase? I've got a
>shopping list of server dream features that require this...
We've built our Web server directly on top of our engine, giving direct
access to our indexes, whether they're built with our tools or our engine
OEM'ed by Lotus, Adobe or whomever. But we're a full-text database, rather
than a relational database -- a rather different animal, with the resulting
performance differences in text v. fielded search and retrieval.
I have to say that it sounds like you're making a bad assumption here --
that a relational database is a good way to store text objects. Certainly
Oracle is making that argument, but I don't believe it.
It seems clear to me that information systems need aspects of each --
relational databases and full-text engines. A database is good for fielded
resource management, including traditional document management functions.
We already have a flat database in our engine for that reason (in addition
to the full-text index). With a key shared between our flat database and a
relational database, the two can be linked together. We have an open
gateway architecture which our clients have used to wire us Sybase, Oracle
etc.
We'd like to integrate the two more thoroughly, as would the database
companies. There are many ways that could happen.
There's a very simple integration of a database and our engine on our demo
pages, accessible from <URL:http://www.verity.com/>. Look at the Photo
Search Demo.
BTW, we're about to hook in a new version to our public pages (we still
haven't announced it as a product, but, um, ah, we keep showing it and
naming pre-release customers) that adds free-text parsing -- it does the
"find me more like this" function.
Nick