I just came across this article:
Are we really ready to call relational database’s “legacy technology”? My first reaction is that column-oriented will only ever be niche software. The article points out that data can be read faster because the columns are together, but what about writing the data? I would think that while data may be accessed by the column, data would be written by the row. So, I would guess that column databases fit well in the data-warehouse niche, but for day-to-day transactional data, we are still going to use relational databases.
Hopefully, I will get the opportunity to work with such a database so that I can have more than just a first reaction.
The most interesting thing to me about the article is the mentioning of BigTable. BigTable is the database that Google uses. Here are some more links on BigTable:
My next question was, I wonder if Google releases the code for BigTable. The best answer that I could find came from Wikipedia: project Hadoop/project HBase.
Hadoop is part of Apache Lucene.