Saturday, January 13, 2007

TupleSoup, Part 1

I am in desperate need of a small embedable database for a project at work. It would be far too easy to just pick one among the many excellent open source Java databases already available (H2, Derby, tinySQL, Mckoi, QED etc.), so naturally I am writing my own. As I work on this project, I intend to regularly post about my design and implementation choices on this blog. The complete code will be posted on sourceforge as soon as it is usable to anyone but me in any way.

If the design and implementation of a small database sounds appealing to you, why not follow me here while I do my very best to document the process in the most entertaining way possible... ok, that might be overstating it a bit - but, if you haven't written one yourself yet, you might want to read along to learn a bit from my mistakes and if you have already written one or two, you might want to stick around to get a good laugh out of those same mistakes. So let me start out by introducing my vision for what this little database should be able to do.

Last year I read Andrew Hitchcock's excellent post about the talk Jeff Dean gave at the University of Washington on Google Bigtable. It was later followed up with an official article (If you have any interest in alternative databases at all, I highly recommend that you read the article, or at least Andrew Hitchcock's blog post). That article really got me thinking about the possibilities of using a simple schemaless database with variable length strings as id's for records for my daily work. I had previously really enjoyed implementing a simple schemaless database, which I have long been thinking about rewriting. So thats where I am at now. TupleSoup will be the core data engine in a new database I am going to start writing (I hope I will also be able to open source that database, but I am currently discussing that with my employer). It will be schemaless and rows will be stored and fetched based on variable length string ids.

I could go on ranting for hours about how I would like to implement this data engine and how I would like to be able to use it, but instead I'm going to stop blabbering now and end this post with a few lines of code demonstrating how I imagine that the finished product would be used.

Table tbl=new Table("datafile",".");

Row row=new Row("kapper");
row.put("name","Kasper J. Jeppesen");


No comments: