BackgroundThe New York Times Company is a media organization focused on creating, collecting and distributing news and information. The Company includes newspapers, digital businesses and investments in paper mills. Its content reaches its audience through its print products, online and through other digital media, including smartphone, tablet and e-reader applications.
ChallengesNew York Times Interactive News Department (IND) consists of technologists that are also journalists, including designers, developers, and reporters. They needed to enable every member of the IND team to keep things online, requiring a reliable path to production and the ability to manage their own code. Moving from Intail to Chef with a Ruby stack they believed could automate enough of the components of Chef to give their journalists a self service tool to deploy applications themselves, scale up a cluster, changing their routing, etc. but ran into challenges with their production environment. The process was slow, complicated to get up and running, checking all scripts from the beginning of time and preventing them from delivering a continuous integration pipeline. They required a new a solution that they could host themselves so they adopted Drone, a Docker based CI system.
SolutionWhen the IND team looked at the underlying components of Drone they discovered the value of Docker. The Docker components had turned what was previously a super onerous provisioning step into a build step, providing a truer sense of versioning. Chef had given them a way to description their infrastructure code but that code was still subject to runtime quarks and if anything went wrong they had to run them again. Docker was able to give them true versioning of the system, when files were modified in the system every change was captured by Docker, not just the ones that were addressed. To keep up with the speed of news the IND team run almost entirely on AWS giving them the ability to scale up 2030 more servers based on the news cycles. Docker’s portability allowed them to easily run on AWS solving their density and speed challenges and delivering a continuous integration pipeline. The IND team is now able to manage their own code, at the speed of news, with the control they require.