Key Principles to Web-Scaling a Network
At the DockerCon 2015 conference today, Docker made good on an earlier promise to extend the reach of Docker containers across distributed networking environments involving multiple hosts.
Scott Johnston, senior vice president of product for Docker, says the addition makes use of software-defined networking (SDN) technology the company gained when it acquired SocketPlane earlier this year. That capability, says Johnston, is critical to make it easier to scale applications built on top of Docker containers that can scale across distributed systems.
However, Johnston notes that in addition to using the SDN software developed by Docker, IT organizations can make use of any SDN software that supports the core networking extensions that Docker has made to its core application programming interface (API). Docker has made that API more accessible by creating a “batteries” plug-in architecture developed primarily in collaboration with ClusterHQ, Glider Labs and Weaveworks.
As IT organizations gear up to deploy more applications based on Docker containers in production environments, Johnston says the ability to deploy orchestration tools across existing and emerging networks will be critical. For its part, Docker today announced that its three orchestration tools, Docker Machine, Docker Compose and Docker Swarm, also now support the network extensions being made to the core Docker API. In addition, Docker announced that these tools, as previously promised, now support the Mesos open source data center operating system platform.
Separately, Docker also announced today that in partnership with The Linux Foundation, it has committed to create an Open Container Project (OCP) standard backed by Amazon Web Services, CoreOS, Docker, Google, HP, IBM, Intel, Microsoft, Red Hat, VMware and Goldman Sachs that at some future date promises to make it easier to deploy application images across multiple types of containers.
While Docker is clearly still bringing all the piece together required to manage Docker containers in production environments, it’s also clear that there will be multiple paths to achieving that goal. Which of those paths winds up being right for any given IT organization may wind up having as much to do with where that organization is starting from today as it does where it hopes to eventually wind up.
