Everyone knows RDBMS, but how about Document-oriented databases?
Consider things that you depart an RDBMS to complete, like indexing, complex query structure, mathematical functions, storage, transactions etc...
I can not say it is the best, however it is quite interesting: Apache CouchDB: http://incubator.apache.org/couchdb/
I've really been researching this myself, and that i haven't began yet, however i intend on using MongoDb for my next project. Other available choices are CouchDb.
Using MongoDb, you're quitting transactions though, so don't apply it any high-value data. (ex. financial)
Some good examples of websites using MongoDb are Github, and Sourceforge
It stores data in BSON format, also it makes a lot sense for web programs.
Another NoSQL options you need to consider is Cassandra (facebook, and digg make use of this)
My first answer, the file system, but you will need to improve your question with increased context.
Not it's a Document Database as a result, but Daisy is effective within this capacity. Mainly since you can author documents there (just like a wiki) you are able to aesthetically look for existing documents. Also whenever you upload exterior files like Ebooks, Paperwork, etc, it fully indexes them so that you can search inside their content.
You may also add custom data fields and contains it's own query language which you can use for a number of tasks. Additionally, it has revision control along with a security/permissions system.
There's a fascinating QCon 2010 presentation by Emil Eifrem where he discusses the 4 various kinds of NoSQL databases one of these he calls the Document - inspired by Notes
- Document databases are KV pairs collected into documents
- CouchDB plenty of credit for quality of Relaxation API and incredibly interesting Erlang
- MongoDB is CouchDB in C++
Accustomed to use Documentum (www.documentum.com) in a large insurance provider I did previously work with. Java based (well the API we used was Java).
Can't imagine it's cheep though.
Usually you store all Metainformation of documents inside a RDBMS. The documents itself are saved in the high end filesystem and therefore are listed in the database. In addition you are able to server your documents using a webserver to third party programs.
Despite the fact that it's fallen from recognition, Lotus Notes/Domino was created for storage of semi-structured data.
Best = most effective/flexible/extensible/scalable/secure?
Notes. It truly has not "fallen from recognition". It simply is commonly under most individuals radar.
Lotus Notes would become qualified as a semi-structured data store - a document driven database - but it is never struck me like a particularly doozy.