Broadcast companies must design the network using a true central-derived model and control the elements within it using SDN routing. It is the most effective way to take advantage of the full potential of IP technology
By Olivier Suard*
Given the increasingly competitive nature of content streaming, any new technology that can improve productivity and enable cost savings will be more than welcome. The ever-increasing adoption of IP technology is therefore no surprise. Considered less expensive in the long run than baseband technology specifically for content streaming, the IP protocol can also make use of existing technology and infrastructure. For many, this represents a huge advantage, as it eliminates the need to make major modifications.
So far so good, except for the fact that this approach can significantly limit the great potential of IP technology. Unfortunately, encouraged by some IP switch vendors, operators sometimes fall into the trap of focusing on products. The primary focus should be on getting the right network architecture and control for the applications and workflows of content streaming.
Three architecture configurations are currently primarily employed in content streaming and other industries.
Centralized star network
This approach is consistent with conventional baseband architecture: all connections go through a large IP router that sits in the main control room. However, this implies that there is no signal aggregation at the network edge: everything must travel to the central router. Fibers must be placed to connect each device to the central router.
To meet the increase in demand and anticipate future needs, streaming companies typically purchase a large router right from the start. The capacity limit is often reached ahead of schedule, at which point the central router needs to be replaced.
In addition, each connected device occupies an expensive high-bandwidth port on the central router, regardless of the actual bandwidth requirement. This raises the cost per port of low-bandwidth devices. The lack of aggregation implies that redundancy must be handled by network edge devices.
Finally, a star network architecture, by its nature, is not ideal for remote locations such as extensions of the primary location, as it implies that all traffic must pass through the central router.
Initially, star networks may seem simple and convenient, but in practice they are very limiting, potentially vulnerable and very expensive.
Central-derivative (spine-leaf)
Large data centers typically adopt a distributed IP network model, known as a spine-leaf architecture. This involves the use of two or more routers in the core (central) and other smaller routers at the edge (derivatives).
By connecting all equipment in each area to derived aggregation routers, and then connecting these to the main routers, companies reduce the number of connections that go directly to the main routers, which simplifies fiber management. If there are fewer fibers, fewer ports are also needed on the central router(s), and a lower cost per port is achieved, especially for devices with low bandwidth.
The central-derived architecture facilitates network redundancy of all devices at a much lower cost. This model also offers optimal flexibility and scalability. Capacity can be increased by adding derived (leaf) or central (spine) routers, instead of replacing existing block routers, as in star networks. Thanks to this, the networks do not need to have a larger size from the beginning, since the capacity can be expanded over time.
While the configuration of a true central-derived architecture may be more complex than others, it is a scalable, resilient and high-performance network structure, which fits perfectly with the needs of transmission companies.
Double star
This third model of architecture is what some might call a spine-leaf, but in reality, it is a "double star" model of architecture. While it requires the use of two central routers, each network-derived router is only connected to one of the central routers.
This solution is not flexible in terms of load distribution and optimization of total network capacity. As the network evolves, it places special requirements on terminal devices that need redundant connections.
Proponents of this architecture often prefer protocol-based automatic routing over SDN (software-defined networking) routing. While a pseudo core-derived network might initially be more attractive due to the simplicity of its configuration, only a true central-derived architecture allows companies to get the most out of their ip infrastructure investment on their premises.
Control
In addition to the network architecture, operators also need to organize and control the IP media network:
Automatic routing
Protocols running conventional IP switches (e.g., IGMP/PIM) allow network elements to make routing decisions based on IP traffic. This means that the decision of how to transport individual media streams over the network can be left to the network, not the operator.
While widely used in IP networks, automatic routing presents disadvantages when it comes to professional real-time media production networks.
For example, automatic routing may not be fast enough to transmit the large number of simultaneous switching events required in live production. You may also have problems with networks where there are loops and where bandwidth management may be poor.
Most of these problems can be solved by IP network experts, but it will be necessary to have their assistance whenever there is a significant change in the configuration, for example, if new studies are added or a large number of cameras are moved from one place to another.
Software-defined network routing
Software-defined network (SDN) routing leaves control of routing to a centralized level of control. The management and orchestration software has a complete view of the available equipment, network infrastructure and services, both those that the organization currently has, and those that are programmed. As a result, you can efficiently make intelligent decisions about routing and flow control. It can also offer the explicit routing feature that streaming companies expect and need.
SDN routing also ensures a higher level of performance compared to automatic routing. The software is also in control of all media flows, which means it has more information and is in a better position to take care of current bandwidth requirements and even requirements that are scheduled. It is even beneficial in terms of protection and security. Orchestration and control software can easily create a variety of pathways to protect against failures, and can also fully control which destination a multicast of a given source can carry, thus reducing security risks.
Unlike automatic routing, SDN routing can, with the right orchestration and control software, easily process any network architecture without risk.
Conclusion
Content streaming companies must design the network using a true spine-leaf model and control the elements within it using SDN routing. It is the most effective way to harness the full potential of IP technology and at the same time helps to obtain an optimal return on investment and gives greater chances of operational success.
*Olivier Suard is Nevion's Vice President of Marketing.
