Planetary-Scale Cloud OS Could Reshape Internet as We Know It
The Internet is a global network that connects most of the planet, but in its most basic form, the Web only connects the world at a networking level, enabling disparate endpoints to connect to servers and to each other.
The Internet today doesn't have a planetary-scale cloud operating system, where applications and policy can move seamlessly from one node to another, across different vendor platforms and stacks. However, that is now changing, thanks to the cloud.
The emergence of a planetary-scale cloud operating system is a notion that Troy Toman, cloud architect at Rackspace, brought up during his OpenStack Summit keynote May 12. It's an idea that relies on a network of OpenStack cloud platforms that are federated throughout the world.
With cloud federation, a user who is authenticated in one cloud can access data and content in another. In the non-cloud federated world, applications and server hosts are mostly stuck in stovepipe silos to a single stack or vendor platform. With a common planetary OS, such silos would not exist because applications and servers could easily move across stacks, vendors and physical locations.
In my opinion, having a common cloud platform like OpenStack for application deployment and then having OpenStack clouds federated with each other is only one part of the total equation to enable a real planetary-scale OS. The other key piece is the ability to seamlessly migrate workloads from one cloud to another, moving applications from one cloud vendor to another, regardless of the underlying technology.
That's the promise that networking giant Cisco Systems is making with its Global InterCloud effort. With InterCloud, Cisco is envisioning that all cloud platforms on the planet can interoperate, where workloads from one cloud can move to another. So if one cloud vendor is running VMware's ESX hypervisor and another is running the open-source kernel-based virtual machine (KVM) hypervisor, it doesn't matter; the workloads and applications can still move.
Going a step further in the vision for a real planetary OS, though, is the need for a truly common network that serves as the foundation. The Internet today provides the basic plumbing, but it doesn't have the same characteristics that server OS administrators tend to have. There is no one easy common way of enforcing quality of service (QoS) or application policy in a consistent way at a planetary scale.
There are now at least two efforts that might end up being complementary and might one day enable the planetary scale for QoS and application policy.
Cisco this past week has been pushing its Application Centric Infrastructure (ACI) message as a way of enabling software-defined networking (SDN) and policy control across a network. If you combine ACI with InterCloud, it could potentially enable control for policy and applications across the planet.
Within OpenStack itself, VMware is pushing its Project Congress effort forward, which could serve as a policy engine for OpenStack clouds. I think it is entirely reasonable to consider that one day Congress, as a higher-level abstraction for policy, could encapsulate Cisco ACI as a source of policy.
As such, OpenStack Congress, or even just Cisco ACI on its own, could one day be the source of truth for network policy and application control—not just for a single network operator or cloud, but for the entire connected planet.
In today's world, network operators, cloud vendors and application providers have to set policies on a more local level within a data center or set of data centers, operated with the same technology.
The promise of the planetary-scale cloud OS is to move up the network stack from a simple layer one of connectivity to a full-fledged application-enabled system, where apps, users and policies are truly global and move at the speed of light.
Sean Michael Kerner is a senior editor at eWEEK and InternetNews.com. Follow him on Twitter @TechJournalist.