Different architectures work for today's new iSCSI video surveillance storage RAID arrays (typically RAID 5), which can be attached from the NVR (or the DVR) or directly to the network (second example).
Photo credit: Photo courtesy Bosch Security Systems
Dr. Bob Banerjee is the product marketing manager for IP video products at Bosch Security Systems, and in this article, he examines new architectures for video surveillance recording.
The topography options of today's IP video recording systems are as diverse as the needs of the organizations using the systems. For a security systems integrator tasked to design the most appropriate and cost-effective solution, the architecture of the system must be the right selection in order to remain suitable for many years to come.
Decisions are based on a variety of factors, some of which include available bandwidth, budget, maintenance requirements, the use of the video, storage requirements as well as the comfort level of depending on the network for the IP video system.
Typically the driving force behind the architecture is where the video will be recorded. This is because this choice forces the key questions of available bandwidth, network reliability, larger shared or smaller dedicated storage, storage reliability and system redundancy. Despite the a (growing) array of possibilities, one of the first major decisions to be made by an integrator during system design is to choose from two major camps: whether to employ a centralized, Network Video Recorder (NVR) solution or a more decentralized approach that used to be addressed by DVRs but now has alternatives including encoders with embedded storage and distributed autonomous iSCSI-based disk arrays. Unlike NVRs, these disk arrays do not use PCs.
Traditionally, NVR-based, centralized recording systems have been favored for a number of reasons, including reducing the cost of storage, which can account for anywhere from 30 to 70 percent of the overall expense of a system. Centralizing the use of storage means the ability to capitalize on economies of scale because hundreds of cameras are able to share highly reliable, large capacity RAID storage arrays.
In addition, locating the recording media and other devices in one environmentally controlled and restricted-access location provides an advantage for maintenance as well as for the security of the equipment itself. Such isolation is common and recommended, including airports, power plants, correctional facilities, education, gaming and government.
Transporting all video to a central location for recording creates great dependency on the network and highlights the main weakness with centralized recording - any interruption in the network means you lose the ability to record video. Not only does this approach demand continuously high bandwidth requirements, it's also reasonable to expect some scheduled, as well as unscheduled downtime, despite the reliability of modern network switches that make an actual LAN failure rare.
Alternatively, using a decentralized system for edge recording can offers many advantages. In the simplest situation, the storage medium -- solid state or hard disk -- is built into the unit. A typical example would be a multichannel encoder, such as a unit with 8 video channels and with up to 1TB of hard disk storage built in to a compact rack unit. At first glance, this may sound reminiscent of the humble DVR, but in actuality we're talking about is a fully equipped encoder with all the capabilities of a pure video over IP device. Now, multiple cameras and encoders can share the storage on a local recording network, while the main LAN or WAN used for viewing and retrieval remains without any recording load. The video does not touch the main network in order to get recorded, which means it is immune from any main network failures.
This approach has been growing in popularity as disk array RAIDs that use iSCSI have been introduced as a better alternative to NVRs, since IP cameras and encoders can stream directly to the disk array without going through a PC first. This system design can often be found in situations where IT is reluctant to deploy PC-based NVR servers â€“ which typically are deployed at the ratio of one NVR for every 64 cameras. The related costs of deploying and maintaining NVR-associated hardware, operating systems and ongoing software updates and anti-virus patches have also added to the appeal of a PC-less approach.
Regardless of the architecture that is appropriate for each IP video deployment, a well-designed project will also feature contingency plans for potential system weaknesses. Distributed architecture systems overcome network down-time simply by its independence from the network for continuous recording. For an NVR-based system, technology exists that provides automatic and consistent replenishment of the network once a network failure has been restored, preventing momentary losses of video from exposure due to wireless transmission. Integrators should look for this technology in conjunction with the ability to record video locally using an internal hard drive or other media (such as a compact flash card) that can fill in the gaps when the encoder was unable to transmit video to the central NVR.
Already, security system manufacturers have begun to introduce video recording managers that distribute video across various iSCSI disk arrays on the network. If one disk array unit fails, the recording manager will simply redirect video data to a backup device. This approach to video management will result in better disk utilization, better load balancing and greater reliability. The throughput will also be greater because of the elimination of the NVR as part of the architecture.
For integrators, the benefits of using a video recording manager are a simpler architecture with fewer items that can fail and less hardware and software for IT personnel to manage. With this type of system, integrators can avoid using NVRs.
A well-designed IP video project will also take into account the combination of solutions in order for maximum flexibility and customization to the system owner's needs. For example, both architectures (centralized NVR and decentralized RAIDs) can be combined. Such a move might pair recording at the edge [such as at for some remote locations which are connected with a low-bandwidth WAN] with other locations that feed video centrally. While each topography has its advantages and drawbacks, today's growing array of solutions gives the integrator more options for system design and improved ability to meet their customers' diverse requirements.
About the author: Dr. Bob Banerjee is the product marketing manager for IP video products at Bosch Security Systems, Inc. He has extensive experience in advanced hybrid analog and digital CCTV solutions and has developed Bosch's IP Resource Center found at www.boschsecurity.us/ip. He can be reached at (717) 735-6637 or firstname.lastname@example.org.