Our fearless leader Jim Falgout has written an article for the Java Developers' Journal on the process he and his team went through while developing a fuzzy matching application. We're proud to note that One Team, One Month, One JVM is the cover article for the March 25th edition.

My favorite part:"The customer had been thinking about the problem for a while and wrote a program in SQL to do the roll-up. The SQL ran in about three hours over 14 million matched pairs. That is way too slow given the matching itself runs in minutes. Realizing the roll-up is a disjoint set problem, we found an algorithm for combining disjoint sets, coded the algorithm in Java and incorporated the code into an operator within DataRush, our dataflow framework. Running against the same data using the DataRush solution on the customer’s same 8-core machine, the runtime dropped to 22 seconds."