It has been nearly a year since I last posted, but I haven't been lazy. I have been busy in my spare time improving the performance of the database operations and adding lots of new features. The relational database table operations are significantly faster now (as are the tag look-ups for Didgets).
A couple of posts ago entitled "Another Piece to the Puzzle" gave the time it used to take to query a 1 million row table with 6 columns at 25 seconds. Now I am able to query a 10 column table (also with a million rows) in under a single second. The time it takes to import a .CSV file has also been greatly reduced.
I can now import a 5 million row, 10 column table in 1 minute and 6 seconds. Most queries against that table now take at most about 2 seconds.
See a video demonstration of queries against the 1 million row table at http://screenr.com/xKmN
Since every column in each table is stored within a pair of Tag Didgets, they each become a separate key/value store. All values are de-duped, so if the column "First Name" has 10,000 rows with the value of "Fred", the actual string value is only stored once with 10,000 references to it. The user can select the Tag Didget containing all the values and view them along with the reference count for each.
Each row in a database table is just a set of values from the separate key/value stores that all map to the same key (e.g. the row number).
Another video demonstration of ways to view all the values in each key/value store is at http://screenr.com/AKmN
We have several additional features that we are working on, but this will give you a taste of how fast our database inserts and queries are so you can compare this against existing database managers.
Remember, we are not running on top of MySQL, SQLite, PostgreSQL, or any other commercial or open source database manager. This is all running just on the Didget Management System. You get all this functionality without needing to install a separate RDBMS.