SDN is the enabling technology of large-scale datacenters. It is part of what makes Amazon, Netflix and Google able to work at scale. Can this technology help us create and maintain large-scale television operation infrastructures? Let’s examine what SDN is, what it does and how it might apply to our industry.
So what is SDN? Essentially, it is an approach to networking that separates the control plane (deciding where network traffic is sent and why) from the data plane (which moves the large volume of packets from here to there). In other words, it uses the network switches to do what they are good at — slinging IP packets at amazing volume — but takes the decision-making out of the switches and puts it up in a control system that has a big-picture understanding of the application. This approach has a significant impact in areas like fault recovery and movement of networked functions from one machine to another (or one datacenter to another).
The SDN technology started as a university research tool, but became commercially important in the large-scale telecom and datacenter industries. SDN enables the application (VOIP call management, Virtual Machine Hosting or large-scale merchant transactions) to steer traffic explicitly to where the application needs it based on actual knowledge, rather than relying on the “convergence” of low-level routing protocols that rarely understand the universe more than a hop away. The smartphone revolution could not have happened as easily — and certainly could not have been as affordable — without the use of SDN techniques in the underlying networks.
Best of both worlds
At Imagine Communications, we are leveraging the technologies and experience developed in the datacenter and telecom industries to re-think television production and distribution infrastructure. The datacenter world was inventing and building businesses and models that were somewhat tolerant of growing pains. In television, the challenge is different — our industry is on the air already, with a very high quality level and excellent operational availability. Any technology transition within the television infrastructure has to improve a scheme that is already quite good, and has to do it in a way that preserves the operating enterprise while getting there.
The television industry also has — importantly — many decades of human operating practices and human-technology interfaces that work, and that align with the organization structures and job descriptions of the people involved. A television operations facility is a complex mixture of people and technology. Even a simple interface like a router control panel has a surprisingly rich feature-set; when someone presses “take” on a router control panel, they expect the route to happen and the panel to reflect a positive confirmation that it happened successfully. We can change the underlying technical process, but the human-interaction paradigm needs to be preserved. It’s not just about human interfaces — there are dozens of subsystems within a typical facility that are interdependent and that use dozens of legacy protocols to communicate with each other; you can change the underlying technology, but the control-protocol behavior needs to stay consistent or other elements will “break.”
The Imagine Communications approach to television production and distribution infrastructure recognizes the importance of maintaining continuity of the human operational roles and tools, and the machine-level integration protocols. It must deliver the economic benefits of flexible IP-based infrastructure and commercial off-the-shelf (COTS) switching hardware, but also integrate with existing television operations facilities and workflows. SDN allows us to leverage datacenter-scale IP switches to carry television signals, while keeping the control of those bitflows up in the routing control system where we can ensure the operational consistency and tie to the legacy integration protocols.
SDN + Orchestration
Another datacenter concept is “Orchestration,” which is the automated construction of a combination of hardware, software and configurations to achieve a result. For example, when you go online and order a domain name and some web hosting for it, an orchestration system generates default pages, allocates a server, spins up a virtual machine to host it and registers the DNS entries to point to it. This happens very fast, and zero human-staff-hours are needed to fulfill your request. In the case of a television operation, orchestration is used for organizing new workflows based on need, by allocating hardware and resources, configuring them to do a job (like setting up network connections between them) and then posting the status of the workflow back to the operator. We have a product called the Magellan™ SDN Orchestrator, which provides familiar “SDI-router” control over IP-connected devices, and also traditional SDI routers. Operators see a standard router control interface, and the operator does not know whether it is an SDI router connection or a 10 Gigabit Ethernet switch and some endpoints getting provisioned. Indeed, at different times, the same button on the same control panel might call one or the other — or a complex connection involving both.
This hybrid approach, integrating the control of SDI and IP equipment together, is important because it allows television facilities to evolve gradually. Big-bang transformations are high risk, and greenfield sites are few and far between. So we have to move step by step into the great future of IP-based flexible facilities. SDN and Orchestration are the critical enabling technologies for functional virtualization and the cloud, managing the signal flows to local and remote computing resources and dynamically scaling “the facility” as needed to best deliver the performance. They are the roadway to leveraging the best advances in IT technology and techniques to drive forward efficiencies in broadcast operations.
With the SDN + Orchestration approach and COTS IP switches, television facilities can participate in the famous “Moore’s Law” capability curve on their core interconnection technology, while maintaining a workflow and control environment that advances based on operational needs. Workflows can then evolve at human/business-driven speeds, no longer dictated by the legacy hardware and able to be optimized for the best business outcomes.