Tracking the Docker PaaS Debate

Now that Docker containers are all the rage among developers basically there are three ways to deploy them in or out of the cloud: on a physical server, on a virtual machine or in a platform-as-a-service (PaaS) environment. In the latter case, a debate has begun over what type of PaaS might be required to efficiently run those containers.

Cloud Foundry is the dominant PaaS environment in use today. Within that open source community, Netflix has created an open source microservices framework, known as NetflixOSS, on which it delivers its streaming network services. Now Pivotal wants to make it simpler to build private clouds using that microservices framework running on the Cloud Foundry PaaS implementation it supports.

Within the beta release of Spring Cloud Services 1.0.0, Pivotal has integrated NetFlixOSS with the widely used Spring programming tools based on Java. At the same time, Spring Cloud Services 1.0.0 signals the unification of the Spring Cloud project and the Cloud Foundry PaaS, said Pieter Humphrey, outbound product manager for Pivotal. As a consequence, Pivotal is making a familiar Java construct available that solution providers can use to build enterprise-class clouds deployed on premise, in a hosting environment or in a public cloud.

With more IT organizations moving to standardize on Cloud Foundry, Humphrey said Pivotal clearly sees demand for deploying Spring applications on an open source PaaS environment that is commercially supported. Integration with NetFlixOSS take that goal a step further by enabling the delivery of microservices using Docker containers more easily within a traditional enterprise environment.

The Cloud Foundry PaaS already supports Docker containers. Support for NetFlix OSS provides a framework for managing those containers. But some proponents of Docker containers contend that microservices delivered using containers require a lighter-weight PaaS than Cloud Foundry. Humphrey scoffed at the idea. He noted that Cloud Foundry was built from the ground up to support a proprietary container that has since then been replaced by Docker.

Solution providers should pay attention to this debate. While clearly there is a lot of momentum behind Cloud Foundry, the rapid rise of containers such as Docker is upending many traditional assumptions about what needs to go into building a cloud platform. While traditional application development models based on frameworks such as Spring are not going anywhere anytime soon, there is now more diversity in terms of how applications get built and deployed. That could mean longer deployment times as IT organizations wait for the debate to settle.