Paying attention to network architecture can keep media networks operating efficiently.
Professional media has special characteristics that must be considered when transporting this content over computer networks. (See the September article “System administration” online at http://broadcastengineering.com/storage_networking/system-administration-0810.) These characteristics can affect everything from router design to disk drive throughput. But nowhere are these characteristics more important than in the layout and configuration of your computer network and in the configuration of key networking components. This month's article will cover networking basics regarding the design and deployment of media networks.
IP over Ethernet is the most ubiquitous networking available by far. Working with professional media, you may encounter other networking technologies such as Fibre Channel, which is used to connect high-performance storage devices, and IP over SONET (Synchronous Optical NETworks), which is used for long-haul video-over-IP applications. However, for this basic article, we'll stick with IP over Ethernet.
I am going to assume that you already have a pretty good knowledge of computer networking basics. If not, there are many excellent basic references available. I happen to like the books published by Cisco.
The focus of this article is to provide practical guidelines for media networks. Here are some points to consider:
Media networks and business networks should not be mixed.
Careful attention should be paid to routers to ensure that network traffic goes where it should and does not go where it should not.
Use the right wire, the right connectors, the right wall plates and the right technique.
Not mixing media networks and business networks
It is important that the traffic on media networks and business networks not be mixed. There are many reasons to keep them separated:
It increases security. Media networks are at the heart of our facility. Keeping office traffic separate from media network traffic increases security and reduces the risk of outages caused by human error.
It allows you to deploy high-capacity networks where they are needed. Media networks move very large files. Keeping the networks separate allows you to deploy high-speed networking where it is needed without having to build out the entire facility using the same transport. You may choose to use 10GigE fiber in the media network but less expensive unshielded twisted pair (UTP) GigE or 100BASE-T to business desktops.
It avoids affecting office network performance with large file transfers on the media network. Even if you deploy high-speed networking technology, the performance of the media network may slow when several clients move large media files at the same time. Some decrease in speed at peak times may be acceptable on the media network, but business office personnel probably will not accept having their systems slow down every afternoon as editors and graphics operators begin preparing for evening newscasts.
It keeps a networking component failure in the office network from propagating into critical on-air operations. I have seen two separate cases where a network card began chattering constantly. In both cases, the network became unusable because of all the traffic generated by the faulty card. Proper network design, including isolating network traffic to business units, will keep a failure in one area from affecting the entire facility.
It follows engineering best practices. There are a number of other reasons it is a good idea to segment networks according to business unit functionality or according to some other method. I do not have enough space to list them all here, but suffice it to say that for many reasons, it is a best practice to segment network traffic by area.
Separate networks isolate traffic
The first step in separating traffic is to create separate networks. You do this both by physically separating the networks (separate cables and hardware, etc.) and by giving them different network addresses. It is possible to run two logically separate networks on the same physical wires, but unless you use a virtual local area network (VLAN), this will have unintended consequences. We will talk more about this below.
Using the simple example shown in Figure 1, the business network is on one physically separate network, and the broadcast core is on another physically separate network. The Internet connects to the facility through a firewall/router, and the two networks are connected through a firewall/router.
You could give both networks the same addresses, but as soon as you tried to allow any communications between the two networks, problems would result. Entire books have been written on network addressing, but for now, there are two types of IP network addresses — public and private.
In a typical network, something called the subnet mask determines which part of the address refers to the network and which part refers to the host or computer. For example, given the network address 10.0.0.0 and a subnet mask of 255.255.255.0, we can determine that the network address is 10.0.0 and that there are 254 addresses available in the range of 10.0.0.1 through 10.0.0.255. In this case, the address is of the form nnn.nnn.nnn.hhh, where nnn.nnn.nnn indicates a network address and hhh indicates a host address. Thus, we could assign business computers addresses starting at 10.0.0.2. (Convention reserves 10.0.0.1 for the network router.)
On the broadcast core, we can choose another network address, say 10.0.1.0, and then begin assigning addresses from 10.0.1.2 to equipment on this network. We have created two separate physical networks with two separate logical network address spaces. But how do you get traffic from one network to another?
Pay careful attention to routers
If you configured the computers in Figure 1, as described above, and connected them all to the same switch, you would quickly find that the business computers could talk to each other and the broadcast core computers could talk to each other, but that business computers could not talk to broadcast core computers (and vice versa). It may seem that we have achieved our objective, but there is a problem. As soon as a media client begins a large transfer, the business network would be affected by that traffic. This presents a challenge. How do we keep these two networks from affecting each other, but allow selected computers on each network to communicate with each other? One way is to use a VLAN. Another, perhaps better way, is to use routers and firewalls. The main purpose of a router is to route traffic from one network to another. The router/firewall in Figure 1 is carefully programmed to allow only particular types of messages from specific computers to communicate across the network boundary. Traffic on the media network is never seen on the office network switch and vice versa. The goal has been achieved.
Continue on next page
Using the right wire, connectors, wall plates and technique
In media networks, it is critical to use the right wire (or fiber), connectors, wall plates and techniques. Anything else will cause network performance to suffer because media networks typically push network performance to its limit. In the limited space available, I cannot go into details, but here are some places where I have seen problems:
Pay special attention to network speed compared to network cable rating, especially in situations where you are reusing existing cabling. You may be able to get away with using Cat 3 cable for 100BASE-T installations (but Cat 5 is ideal), or you may find that Cat 5cable works for GigE (you need to be using Cat 5E or Cat 6). Over time, the network may become unstable.
Watch out for patch panels or patch cords that are not properly rated for the network speed being used. I once spent several days troubleshooting an intermittent problem affecting an on-air automation system only to find that the patch cords being used in one part of the system were flat ribbon cable and were not rated for use in Ethernet networks at all.
Use the proper cable-laying techniques, especially if fiber is being used. If you exceed the maximum bend radius of a fiber, the light cannot follow the fiber, and errors or link failure will result.
Watch out for maximum length cable runs and maximum overall network length. These specifications are required due to tightly controlled timing on Ethernet networks. If transit time across the network becomes too long, errors will occur regardless of the quality of the cable.
Pay attention to instructions and the quality of workmanship if you are terminating cables or fibers yourself. Get a network cable tester, and test every cable. Cables can look fine but may not meet network specifications.
Brad Gilmer is president of Gilmer & Associates and executive director of the Advanced Media Workflow Association.
Send questions and comments to: firstname.lastname@example.org