Pods, Orchestration and Scale

Our first generation of data center switching products brought together SDN and photonic switching in support of an application-driven movement we call Affinity Networking. Our central thesis has long been that our industry has to collectively move beyond what has been a static networking environment. The challenges of managing edge policy have only gotten worse as the infrastructure around the network has become more fluid. It stands to reason that the network simply had to transition as well, from static to responsive.

The notion that the network needs to be more of a willing partner in delivering application experience is not necessarily unique. But solutions to delivering such an infrastructure have hinged primarily on management, ignoring the very static nature of the underlying connectivity the network provides. Plexxi has embarked on solving both the management issues (through its SDN controller) and the static nature of the underlying capacity (through its photonic switching capabilities).

Over the past several weeks, we have been revealing elements of our second generation of switching products. In summary, they include:

• Pod Switch Interconnect — We have taken our Affinity Networking solution and packaged it in a form factor designed explicitly for pod deployments. The fixed-configuration pods come in 6, 12, and 36-switch sizes. The switches are cabled together through a passive optical chassis that provides rich, dynamic capacity allocation. The fixed configuration leverages cost-effective components, allowing us to deliver a a pod solution at very affordable prices without sacrificing any of the Plexxi-ness behind our application-driven infrastructure.
• Data Services Engine — The currency of integration and DevOps is data. Automation is not the act of replacing keystrokes with scripts but rather the diligent orchestration of activities across the whole of the IT ecosystem. To that end, Plexxi has developed the Data Services Engine. It acts as a central communication point for infrastructure data, allowing disparate sources of information to publish and subscribe to heterogeneous data sources. However, integration is not a vendor-specific requirement, so Plexxi has developed the Data Services Engine in a vendor-agnostic way and committed to contributing it to an open source community.

These offerings are important building blocks to help deliver an application-driven experience. The dynamism demanded by even current-generation applications requires a physical network whose capacity is fluid. Managing capacity effectively can only happen if there is a global view of the network as a resource, making a controller critical. But capability — no matter how powerful — is useless in isolation. So pairing this with an open data services engine and an application abstraction model is absolutely critical.

Our path toward an application-driven, highly-orchestrated, and deeply integrated IT infrastructure began more than three years ago. This week, we take the next important step.

For more information on these elements of the Plexxi solution, check these additional resources:

• Pod Switch Interconnect on Slideshare
• Data Services Engine video from Networking Field Day 6
• PSI video from Networking Field Day 6

​- See more at: http://www.plexxi.com/2013/11/pods-orchestration-scale/#sthash.IsW63ycC.dpuf