It's Friday afternoon. You're white-knuckling a stress ball while your node script produces that data analysis your boss says is a 'must have' by day's end. Margarita dreams fade with each idle tick of unused cores. With happy hour on the line, squeezing every ounce of power out of a machine is a must. In this talk, we'll use node's 'cluster' and 'child_process' modules, and some clever planning, to tightly pack long-running jobs into an efficient schedule. We'll discuss iterative improvement and efficient parallelization in a distributed system. Finally, we'll try our scheduler out and, ultimately, get happy hour back.
Cabell Maddux, Software Engineer, QxMD Software
Location: Level 1, Fisher East