IT trends impacting IP video: The network's bandwidth

You're not alone if you believe that even a small network surveillance camera system could overwhelm a company's bandwidth, making it an ineffective pipeline for security. Despite evidence to the contrary -- such as installations where thousands of video channels stream over a single network -- confusion and apprehension often dominate common sense when it comes to separating fact from fiction regarding bandwidth consumption. Part of the problem stems from the variables used to calculate bandwidth specifically for surveillance video alongside performance levels of the networking equipment we install.

To provide clarity, this article explores the variables affecting bandwidth consumption, including ways to employ some of the latest advances in Ethernet standards to help both security and IT professionals make intelligent decisions regarding video surveillance installation choices.

Getting to the source of bandwidth misconceptions
At Interop 2011, the premiere technology tradeshow for the networking industry, it was obvious to me that most networking professionals believe that video on the network consumes far too much bandwidth. So, when security professionals recommend deploying higher quality, IP-based surveillance solutions, IT departments immediately refuse to support the project. Looking at the performance levels of switching equipment as well as the bandwidth consumption of today's networked surveillance cameras, one wonders how they could have reached this conclusion. So I asked them.

It turns out that their perception was based on external sources such as YouTube, and other sites that deliver downloadable content. Network professionals cringe at events like the recent Royal Wedding and basketball's March Madness tournament because they know hordes of employees will stream live video or download highlights. The operative word is "download," which points to a very different technical reality than what is created by surveillance content.

A company's connection to the Internet is typically smaller than the bandwidth available across its local area network (LAN) – usually at least 10 times smaller. What IT is really concerned about is the download speed from the Internet, which typically has nothing to do with internal video surveillance activity. From their perspective, it's a budget issue of how big a pipeline they can afford. In contrast to hordes of employees simultaneously downloading content from the Web, surveillance video is a LAN activity viewed by a limited group of authorized viewers and therefore shouldn't cause IT professionals anxiety about overwhelming the pipeline.

Calculating network requirements for video surveillance
Scoping out the bandwidth requirements for a surveillance system begins with defining your viewing and recording needs. These viewing and recording requirements differ in many cases, often resulting in dual-streaming solutions that factor significantly into the overall amount of bandwidth consumed. In each case you need to specify the resolution and the images per second to be viewed and/or recorded. With a few exceptions, the de facto standard for compression has become H.264. But the bit rate will fluctuate depending on the amount of motion in the scene.

To help you figure out how much bandwidth you might need, let me walk through a scenario using the bandwidth calculator from my employer Axis Communications. That calculator can be found online at

The scene I've chosen contains hundreds of people exiting a train platform during rush hour. The video is to be viewed and recorded in 720p which amounts to 1280x720 resolution. I've selected 18 frames per second for both viewing and recording. To put it into perspective, this is the same viewing experience that most of us see while watching the Super Bowl. Using advanced H.264 compression, I've calculated a total bandwidth consumption of 8 megabits per second (Mbps) which is derived by adding the viewing and recording bandwidths together. Assuming the camera is attached to a 100 Mpbs network switch, this means only eight percent of the total available bandwidth will be dedicated to that camera.

If you are viewing 50 cameras using this configuration, the total bandwidth consumption would be 400 Mbps, which represents 40 percent of a 1 Gigabit per second (Gbps) switch port. So the math proves that LANs can easily handle the demands generated by networked surveillance video with bandwidth to spare, given that most surveillance installations don't view video 24/7 and generally run at six frames per second or less.

Keep in mind that the source of the video traffic and the destination are static – meaning that the camera is always sending video to both the viewing station and the recording platform attached to the 1 Gbps switch port. This essentially segments that bandwidth away from any other network traffic, such as e-mail, which is generated in daily business operations. Other departments would not be given the security credentials to access the surveillance video and thus it would never impact their share of the network bandwidth.

When to be concerned about bandwidth saturation
So are there any situations where surveillance video would cause IT professionals some concern? If you were downloading video across a wide area network connection, then you'd need to do some proper planning and budgeting. For instance, take a company that has a dozen remote offices each with two cameras that need to be recorded and viewed at headquarters. If you maintained the same configuration as our earlier example, each site would require upload speeds of 16 Mbps just to handle the generated video, let alone any other business applications. The headquarters site would need a download connection speed of more than 200 Mbps to record and view the 24 cameras across all 12 remote sites. Although it is technically possible, for most companies it would not be economically feasible.

In this situation most consultants would design a system that proposes localized storage at the stated video quality with a reduced resolution and frame rate for the viewing platform at headquarters. This compromise lets users view events in real time and download the higher-quality video only for those events that require an increased level of scrutiny. Implementing this design allows the network to handle a much lower sustained throughput with only sporadic bursts of increased bandwidth.

So what's the 'Tech Trend' from this column?
The trend is in applying new Ethernet standards to expand the pipeline. With the IEEE ratification of 40 and 100 Gbps Ethernet standards in June of 2010, we're now entering a whole new level of opportunity for aggregating video system bandwidths. The standard, IEEE 802.3ba, defines achieving these speeds by combining multiple 10 Gbps channels. At 10 Gbps, we can deliver more than 1,500 HDTV 720p video channels over a single cable (Note: HDTV standard means full frame rate, widescreen aspect ratio, guaranteed color fidelity, and 720p or 1080p/i resolution). Multiply that by 4 or 10 times and you start to understand the capabilities of the Ethernet as a transmission medium. Is it practical to envision 1,500 HDTV video channels across one cable? Yes, if you think about the typical camera counts in municipal monitoring, transportation hubs, and critical infrastructure. With the expanded Ethernet capability, camera counts soaring into tens of thousands of recorded video channels operating in extremely high quality will become commonplace.

Opening the bandwidth dialogue between security and IT
It is clear that bandwidth consumption is a much maligned and misunderstood topic. Security professionals need to truly understand how consumption is calculated to have an intelligent conversation with their IT counterparts. As I have learned, IT professionals also need to understand the nature of network surveillance video as a data source that is not necessarily generated and downloaded from external sites. Whichever camp you are in, blithely dismissing the topic won't cause it to disappear. In the face of existing installations already scaling into the tens of thousands of network video channels, it's obvious you can present a convincing argument that network video surveillance won't drain a typical LAN.

About the author: James Marcella has been a technologist in the security and IT industries for more than 17 years. Besides writing a quarterly column for SecurityInfoWatch about technology trends which affect the surveillance industry, he also serves as director of technical services for video surveillance manufacturer Axis Communications.