BackgroundGeneral Electric (Appliances) builds innovative and energy-efficient appliances to help improve the lives of their customers. The division of GE has 12,000 employees and a revenue of more than $12 billion per year.
ChallengesGE Appliances' initial application development process was labor intensive, taking an average of 6 weeks time to move from development to production due to mistakes, reviews and rework of application configurations. This process lacked repeatability due to its manual nature. With several projects underway, like implementing their new self service web application, and migrating a 61 year old legacy data base to their private/public cloud environments, they needed an easy to use solution that could enable their application owners to be agile.
They decided to move to a cloud environment, thinking that it would solve the issue of time from dev to prod, but even after moving to cloud their process still took up to 3 weeks, lacking the repeatability that the company sought. GE Appliances also started looking at different IaaS vendors and started out using Puppet. The theory was that developers would develop puppet rules and then continue to develop these rules over time. Unfortunately, they experienced a poor rate of adoption of Puppet, as app owners disliked the high barrier to entry.
SolutionThe ease of use and high portability of Docker containers gave GE the ability to build and run in any environment. Docker solutions offered a much lower barrier to entry, compared to some of the other solutions they were looked at. Today GE appliances uses Docker containers as the basis for their self service application called Voyager. Bridging the self service Gap, Voyager is a new web application that they developed. With it, the company can now write a Docker file, access the Voyager UI to make a request, and then the service discovery layer can quickly notify them of a running service.
In the past they used VMware, which meant running just one application in a VM, but by leveraging Docker instead they now have on average 14 applications per container, a 14x improvement, giving them a much higher infrastructure density. With Docker, GE is able to support both their 30 year old legacy applications, as well as the newer applications they are running in the exact same environment. To top it off, GE embarked on a journey to migrate everything out of a 61 year old datacenter to their new hybrid cloud environment, a project that they estimated would take years. After using Docker for only 4 months, they have already migrated over 60% of the legacy datacenter.