All 14 articles in this series are now available in our free eBook ‘Understanding IP Production Networks - 2026 Edition’ – download it HERE.

All articles are also available individually:

Network timing requirements, differing working practices and protocols, and data integrity all help to compromise communication between broadcast and IT engineers.

Timing in broadcast is tightly defined, and a thorough understanding of legacy television systems is required. IT engineers use asynchronous full duplex systems, expect there to be network failure, and use protocols that effectively slow transmission to make sure data has been accurately delivered. Broadcasters use synchronous one direction connectivity and assume the network is as robust and reliable as SDI.

In this series, we look at networks from a broadcast engineer’s point of view, giving a better understanding of core IT concepts and enabling them to communicate with colleagues in the IT department.

Network Basics

To fully understand IT networks, we must understand the problem we are trying to solve; a network is needed to allow users to exchange data predictably, reliably and securely, and provide control of one computer over another. This is true of PC’s, servers, IP-camera’s, production switchers and control panels, and the more secure and reliable a system, the more complex it becomes.

A network must be resilient, fast, and reliable to give the best user experience. To explain the roles of routers and switches we start with a basic network of four PC’s and two servers connected in a simple IP over Ethernet network using CAT5.

Ethernet has three forms of physical interface: coaxial, twisted pair and fiber optic. They all send the same type of packets of data but differ in their duplex as twisted pair can send and receive data at the same time, but coaxial and simple fiber optic cannot. Transmission speeds are faster on fiber optic and coaxial.

Few computers use coaxial connectivity as twisted pair is cheaper and more robust. Fiber optic tends to be reserved for high bandwidth switch and router connection due to its higher cost and fragility. 

A hub with twisted pair infrastructure (CAT5) could be used in a simple network. The hub is like a distribution amplifier allowing mapping of one-to-many transmit and receive pairs. The hub has no intelligence and will route a packet received on one port directly to all its other ports.

In a hub network, security becomes a problem as all users would be able to see data being exchanged between each other’s computers and servers.  For example, all users would receive transactions associated with the finance server.

Computer network cards receive all datagrams on the connected network and will usually discard those not intended for them. With the right software, it’s easy to decode the datagram and view restricted and sensitive financial transactions. This is true of all the systems running on any of the servers.

Lost Packets

Lost packets of data occur as network traffic increases and the physical switch and router links become quickly saturated, and this is further exasperated by micro-bursts of data that can overflow egress buffers. Protocols such as TCP (Transmission Control Protocol) can remedy lost packets but do so at the expense of increased and variable latency. This is one of the reasons standards such as SMPTE ST2110 uses UDP (User Datagram Protocol) as it operates a fire-and-forget transmission system, resulting in predictable and low latency. However, when using UDP, lost packets cannot be easily recovered.

Ethernet is a packet switched system; each PC will monitor the transmit line and wait for a gap so it can send its own packet. Although the packets are of a fixed size, the frequency with which they are sent is random across all the connected computers on the network. Another computer may be listening at the same time waiting for the same space, and two or more computers could try and simultaneously access the transmit pair resulting in a collision and packet loss, and slow response for the user.

A network router or switch using point-to-point connectivity will protect against collisions and congestion and is one of the reasons routers and switches are used, other reasons are to provide resilience and security. Switches send packets at the Ethernet packet level (layer-2) and routers route packets at the Internet Protocol level (layer-3).

A layer-2 Ethernet network can be thought of as an isolated network that keeps collisions to a minimum and reduces saturation of a network by keeping most of the datagrams within the locality of the network’s neighbourhood. Multiple layer-2 Ethernet networks are then connected using either a layer-2 bridge, or a layer-3 gateway router. Only datagrams that need to be sent between the different layer-2 networks travel across the bridge or router, thus keeping traffic to a minimum.

In the ISO seven-layer model, IP packets are encapsulated by the layer 2 Ethernet frames. This might seem like an unnecessary overhead; however, the IP protocol is independent of the transmission network and abstracts the data away from the hardware limitations of Ethernet. It’s entirely possible, during the lifetime of an IP packet, that it will be routed over non-Ethernet networks such as ATM (asynchronous transfer mode) or WiFi. With IP, we need not be too concerned with the medium the data is travelling on. In this context the IP routing is acting as a gateway between different network physical layers.

Each Ethernet card in a PC or IP-camera has its own unique hard coded address called the media access control (MAC address). Each camera can be configured to have a unique IP address, so a faulty camera C1 can be replaced with the same IP address. The MAC address will have changed but the address resolution protocol (ARP) in the routers would detect this and reconfigure themselves. 

Managed Ethernet switches provide a better solution but have limited capability. The switch is configured with the MAC address of each computer connected to its ports and will send traffic only intended for the associated computer thus reducing network traffic on each connection. For these reasons Ethernet switches tend to be used in fixed high-speed applications such as core network switches and head of rack topologies. They are faster as there is less information to process in the Ethernet datagram header compared to an IP header. For example, there is no “time to live” value to be updated.

Resilience & Security

IP addressing schemes offer greater flexibility and allow administrators to specify their own IP number schemes. Security is improved as routers can be configured to make sure finance transactions only go to authorized computers and IP-camera’s only send their pictures to monitors and production switchers in the studio. Programs such as Ping can be blocked to stop hackers detecting computers and attacking them through DoS (Denial of Service) attacks.

Automatic routing algorithms provide resilience by detecting a broken link and sending the data via a different route. Multi-path links can be used between studios and outside broadcast units consisting of different types of media such as fiber optic and satellite. Users are unaware that routers have switched to a different path when a link breaks.

One interesting aspect of IP in broadcasting is that of signal routing.  In traditional broadcasting, we use point-to-point distribution through SDI signal transfer. Here a camera sends a video signal directly to production switcher with guaranteed latency and bandwidth. A Destination Address (DA) may be used to provide one-to-many mapping, but this still guarantees latency and bandwidth. Essentially, SDI is a synchronous distribution system, and it is this synchronicity between sender and receiver, as well as in the underlying network that provides the reliability and low latency broadcast engineers have come to expect.

IP networks are diametrically opposite to synchronous networks as they are asynchronous by design and packet switched. This provides far greater flexibility, but at the expense of variable and indeterminate latency, and compromised data integrity. We can fix the data integrity by using TCP, but the variable and indeterminate latency is still a challenge. It is possible to fix this, but in doing so we end up with a network that lacks flexibility and we quickly fall foul of the rigidity that is inherent in SDI.

There is always a compromise; DSI networks have guaranteed latency and bandwidth, but at the expense of rigidity and a lack of flexibility in the design.. For example, it’s almost impossible to send an AVC encoded media stream over SDI. However, with IP networks, we gain significantly more flexibility (we can easily send uncompressed video and AVC encoded media streams over an IP network), but we do suffer from packet loss and variable latency. Consequently, exchanging video and audio signals reliably in an IP network can be challenging.

Routing Challenges

As well as all the timing issues that are addressed in the subsequent chapters of this series, we also have the challenges around signal routing. In SDI networks, our point-to-point network gives a clear indication of the sources and destinations of the signal path. Mnemonics on SDI routers clearly show where a signal has come from, and where it is going to. 

Following cables by physically tracing them or comparing cable numbers to the facility schematics, makes understanding the network topology relatively straightforward. However, the same is not true of IP networks.

Multiple source and destination devices share the same links between network switches and routers, and in the case of resilient networks, which most are, we have no guarantee of the route IP datagrams are taking, especially if network routing protocols such as Equal-Cost Multi-Path (ECMP) are employed. We will look at ECMP later in this series.

Understanding an IP network topology can be challenging, especially in dynamic networks which route over different links. To address this, IT systems have adopted SDNs (Software Defined Networks), where the data and control planes are processed independently. This is similar how SDI and AES networks use SDI routers to switch the SDI and AES signals from a central control panel.

In a broadcast network, there may be many thousands, and even tens of thousands of IP addresses describing the device endpoints around a network. Keeping a record of these is a logistical nightmare. This is not new to IT, and they have designed systems such as discovery-networks using DNS servers that map IP addresses to human-readable names. Audio engineers have adopted solutions using Session Description Protocol (SDP) to describe signal metadata such as sampling rate and bit depth.

Broadcasters are now working to develop these ideas and open systems such as AMWA’s NMOS to provide interoperability, discovery and a central database of signal routing, formats, sources and destinations, thus reducing the need to keep reams of spreadsheets that need manual intervention to keep them up to date.

IP networks have much to offer broadcasters in terms of flexibility and scalability, but we need to take a deep dive into how IP packets are distributed in networks to truly leverage their power.