NYTimes uses Hadoop, S3, EC2, and some custom code to handle PDF generation for 4TB worth of data | Heykuki News