As I travel around the country, I find that more and more broadcasters are deploying IT-based technology in the core of their facilities. But at a recent SMPTE meeting in Atlanta, people expressed a high degree of frustration with IT engineers, while simultaneously saying that they recognize that they need the skills that the IT engineers possess.
As IT-based technology becomes more central to our core operations, one area of extreme importance is switching. Analog and serial digital interface routers are critical to our operations, and broadcast engineers understand the nuances of these components better than anyone, except perhaps the people who designed them. The time is soon approaching when broadcast engineers will need to understand Ethernet and Fibre channel switches just as well.
For those of you who are saying, “Ethernet switches in my core!? No way,” I would like to remind you that the same thing was said about MPEG compression, DV cameras for news and countless other technologies.
Not only have many of us had to deal with the introduction of new technologies, but also we have had to implement these technologies under closer financial scrutiny than ever before. In this environment, we often face questions about why we need to purchase a $2000 Ethernet switch when the same switch is available for $95 at a local office supply store. Well, here are a few areas to consider:
The ability of a switch to get all the traffic from input to output is affected by the overall backbone capacity of the switch. If both switches have 10 ports, the less expensive switch may have a backbone capacity of 200Mb, and the $2000 switch has a backbone capacity of 1Gb. This means that theoretically, the more expensive switch will be able to carry all the traffic without delaying or dropping traffic because of insufficient bandwidth on the backbone.
The more expensive switch is non-blocking. That is, it can carry the maximum amount of traffic that could possibly be presented at its input without dropping any traffic. However, when the less expensive switch is presented with data to switch at all of its inputs, some of the data will have to wait because the switch does not have enough capacity to handle it all. This is known as blocking.
Because of the way Ethernet and various protocols are designed, a blocking switch may result in slower throughput, but it is not usually fatal to the network. In other words, a non-blocking switch will make the overall network faster, but a blocking switch just means that the network will run slower. Depending on the criticality of the network, a loss in speed may be acceptable. But if the speed of delivery of data across the network is critical, then a non-blocking switch may be worth the extra money.
More expensive switches frequently provide remote management. Remote management allows the people responsible for the network to be able to collect performance and maintenance statistics from the network components while the network is in operation. This is usually provided using the Simple Network Management Protocol (SNMP).
Typical SNMP applications allow the user to build a map of the network. The components of the network can be color coded to show their status. Such a map provides a quick assessment of network health and can allow maintenance personnel to quickly determine the location and nature of a fault. Some SNMP monitoring applications provide phone-home capability in which the monitoring application notifies maintenance personnel of failure by e-mail or cell phone.
More expensive network switch components can be configured to provide redundancy on several different levels. Port redundancy allows the system to mirror traffic on multiple ports. Should one port fail, the second port continues to function. Of course, this capability assumes that the server, workstation or other equipment connected to the network is capable of using two network interfaces and of switching to a backup interface should the main interface fail. While this capability may seem rather exotic, I am finding that it is becoming more common.
If your network is only supporting conventional traffic such as e-mail, network file services, Web browsing and so on, chances are that the less expensive switch will perform just as well as the more expensive model. However, if you are trying to do multicast streaming or QoS reservation, you will most likely find that the switch will not meet your needs. The reason is that as users needs become more advanced, more protocols have been developed. Some of these protocols build intelligence into the network itself.
In the case of multicast, there is a lot of knowledge built in to the network about who is participating in the multicast and where streams of the multicast are replicated. With QoS reservation, there are a number of protocols available. There are also small variations in implementation of these protocols.
Even though the Internet Engineering Task Force (IETF) has produced a number of Request for Comment documents dealing with QoS, the implementation of these in less expensive routers is spotty at best. The result is that a client requesting a specific QoS may not receive what it is expecting because a less expensive switch somewhere along the way did not know how to interpret the client's request.
If your needs are basic, then there is almost no need for customer support on a switch in an Ethernet network. In this case, the less expensive switch is the way to go.
One should realize that these switches are produced in consumer quantities. It is almost impossible for a manufacturer to provide in-depth customer support on a product when there are hundreds of thousands of items in the field. If you are using more advanced features, such as network management or less common protocols, then you may find that you have a need for more advanced customer support.
In many cases, it is not necessary to have a switch that is software-upgradable. But if you are deploying the switch in a network where you need to be able to deal with advanced protocols or where the switch is also serving as a firewall, then software upgrades will be important.
Usually upgrades are downloaded from a Web site and then loaded on to the switch. When the upgrade is executed, new software is loaded into EPROM on the switch. In all but the most expensive switches, you will have to reboot the switch to load the new software.
Layer 2 level security
Security on broadcast networks is frequently a serious concern. In most facilities, the last thing you want is to have someone carry an unauthorized laptop into a technical area and plug it into the network.
South Figure1. Layer 2 security allows the network engineer to permit traffic to pass through a switch based upon the MAC address of the individual workstation. Click here to see an enlarged diagram.
More expensive switches can be programmed to allow only certain computers to connect to the network based upon that computer's Media Access Control (MAC) address. Once a computer's MAC address is verified, the computer's packets are allowed to transit onto the network. While MAC addresses can be spoofed, this approach does provide an enhanced degree of security and it prevents someone from plugging into the network inadvertently. (See Figure 1.)
Broadcasters frequently would like to prevent particular protocols on their networks. While undesired traffic is usually blocked at a firewall, it is possible to program a more expensive switch to refuse to pass particular protocols. For example, you could block the SNMP if you wanted to be sure that mail traffic was not allowed on the network.
One key thing to consider is that you can mix full featured switches with less capable ones, but you need to be careful. For example, if you are using QoS protocols that are not supported by the less capable switches, then QoS will fail across the entire path. This points out the value of becoming educated about IT. Learning about the technology will allow you to take advantage of high-volume, low-cost products to lower your overall facility costs. But you must know what you're doing to get satisfactory results.
Brad Gilmer is president of Gilmer & Associates, executive director of the AAF Association and executive director of the Video Services Forum.
Send questions and comments to:firstname.lastname@example.org