- The network is very key in delivering usable IP video.
- Working with an end user's IT department is pretty much a must.
To determine the ideal video recording and storage solution, the number of channels is not the only variable; it is just an approximate guide for computing and network requirements and solution design. Other factors include picture resolution, acceptable video compression, required frame rates, data traffic, network layout and single point of failure, said Stephan Beckmann, Video Product Marketing Manager for EMEA, American Dynamics (a Tyco Security Products company).
“One camera may deliver a small picture in a small resolution, low frame rate and some compression. That camera may give us a data stream of 20 Kbps,” Beckmann explained. “Now, a second camera, with a full HD resolution, a higher frame rate (say, at 25 fps) and low compression, can easily give us an 8-Mbps data stream. The system design has to focus on the expected amount of data delivered by the cameras.” Rather than trusting spec sheets, tests will help determine whether processors are actually able to handle the amount of data coming from cameras. “A system of 64 cameras can produce a 400-Mbps data stream that requires a very powerful hardware platform, and also some dedicated testing to see if the NVR really can process the data.”
The focus is usually on processing incoming streams of data when selecting or testing NVRs and servers. Since NVRs need to route data to other storage platforms, output processing power needs to be evaluated as well. “If the NVR can receive 50 Mbps from the cameras, can it also stream the same amount of data to the monitoring client?” Beckmann asked. “Some NVRs in the market are limited on the output side. You may have to use more NVR units because you have to deliver a certain amount of video to the clients.”
The NVRs or servers may be able to process, but then the network may not be able to send the data back to the storage network, said Tom Larson, Director of Global Accounts, BCDVideo. “The network is very key in delivering usable IP video — not so much on small jobs, but definitely on the larger-scale jobs. Larger scale means higher bandwidth requirements.”
As such, the significance of network management is now also on the rise: a camera network and an operational network. In some bigger projects, operational networks are split further into cameras and NVRs (front end) and NVRs and storage (back end), Beckmann said. “This approach has a few advantages: all cameras are protected in their own network; cameras and operational IT equipment cannot influence each other; IT maintenance on the operational network does not affect video recording; and as the NVR is the only gate into the camera network, user access management for video is simplified,” Beckmann said.
The operational/client network is sometimes shared with the end user's corporate network, if the end user requires the ability to go to any computer and pull up video. “You're taking up no bandwidth on that network, unless somebody goes to their computer and pulls up video,” Larson said. “It results in much better control on that corporate network.”
If the cameras are dispersed, the network environment may also be affected by the number and brands of switches in a network. “If you circle through many switches, your video information must be transmitted through many levels before it can reach your NVR,” said Andrew Yu, Security PM at Qnap Systems. In theory, routing through many switches is not a problem; in practice, many switches do not deliver the required performance. Especially when transmitting megapixel data, large data streams may yield lagged images rather than smooth streaming video.
Regardless, working with an end user's IT department is pretty much a must. Unknown factors on a corporate IT/IP network, including bandwidth, network topography and latency, could impact a surveillance system. “There are a lot of factors that the integrator doesn't control when the network is being used for other things,” Larson said. “When the network belongs to the IT department, the integrator can't just go in and fix it because it's not his network. This is where you might have problems: the IT/network guys can sometimes be uncooperative with security integrators.”