Last week was a big one for Docker and the container space. Docker made some exciting product announcements at DockerCon Europe. Simultaneously, CoreOS announced it’s effort to define an Open Container Format in the form of ‘App Containers’ and Rocket. The community responded with a strong reaction. Some raised questions about the benefit of alternate container engine instead of attempting to ‘fix’ Docker. Others expressed their support for Rocket and CoreOS’s effort around keeping the containers simple and standardizing the container specification.
Conflicts in the community aside, I believe that introduction of the App Container specification is a great thing for the end users. This reminds me of VMware circa 2007 when the company embarked on a pledge to define an Open Virtualization Format (OVF), in an attempt to unify the diverse virtual machine types created by vendors such as VMware, Microsoft and XenSource under a single unifying specification.
Unfortunately though, VMware’s efforts around a standardized specification for virtual machines, while received favorably for the most part, did not reach the level of success the company intended it to. The result is a fundamental lack of portability, which has been a pain-point for customers and has lead to creation of multiple islands of infrastructure. The importance of a standardized specification for virtual machines is critical today more so than before, with multiple virtualization platforms getting closer to parity with one another. At Platform9, one of the core propositions we make to our customers is the single pane of glass management interface that spans across different hypervisor types such as VMware and KVM. OpenStack fundamentally enables embracing such diversity under a unified interface. However, not having a standardized virtual machine specification imposes many operational challenges that our customers face today around conversion of virtual machine Images and workloads from one type of hypervisor to another.
So, while some may view the emergence of ‘App Containers’ and Rocket as just a competitive announcement, there is a real customer benefit at play. Customers should be able to choose container technologies without fear of vendor lock-in, and use the management systems provided by vendors on the basis of their unique management capabilities. And that is what a widely adopted standard for containers could bring about. Without such a standard, one of the core benefits of containers: portability, would be compromised and users would have to deal with the hassle of translation across different systems.
At Platform9, we believe strongly that hybrid clouds are the future. And containers are a key ingredient to this future. The benefits containers offer in the form of lightweight model, mobility and portability are certainly unmatched by Virtual Machines or any other technology today. This however means that it’s critical for containers to remain a fundamentally portable unit of technology, and for them to evolve past their current limitations around security. Standardization is key.