Factual processes nontrivial amounts of data. Our analyses may range over 1011 records, reading hundreds of gigabytes to hundreds of terabytes of source data and intermediate representations. At this scale, performance optimizations can save us significant time and money. We use VisualVM, jhat, and Yourkit for memory and CPU profiling, and the excellent Criterium for microbenchmarks in Clojure. Th
![Factual | Profiling Hadoop Jobs With Riemann](https://cdn-ak-scissors.b.st-hatena.com/image/square/815ba4a65e3a57d6617bb3c76315098cb5e92340/height=288;version=1;width=512/https%3A%2F%2Fwww.factual.com%2Fwp-content%2Fuploads%2F2018%2F01%2F2014-08-12_display.png)