One of the most important aspects involved in the design of a video surveillance solution involves storage: size matters, but the choice of the right hardware is also important.
One of the most important aspects involved in the design of a video surveillance solution involves storage. Apart from storage size, the choice of the right hardware is also important.
“Most people think all storage is the same, but it’s definitely not. And the vast majority of storage technologies were designed and built for a different type of application –not video surveillance,” said Brandon Reich, Senior Director of Surveillance Solutions at Pivot3.
How Video Surveillance Data Differs
“There is a common misconception that video data, being digital, are just like any other data. In reality there are important and crucial differences between video data and ‘general’ IT data and such differences are amplified in large-scale surveillance systems,” agreed Alastair McLeod, Group CEO of
Veracity UK. These differences have substantial implications on optimal system design, lifetime and TCO.
“The differences lie not so much in the data themselves, but the way in which they are generated, stored, retrieved, modified, manipulated, presented and deleted,” said McLeod.
In most general purpose IT storage and retrieval systems, the data set is of limited size as compared with video and is added to incrementally at a very modest pace. The systems are indexed and optimized for very fast searches across the database. Typical data operations read the data, modify it in some way and write it back. Thus there are a lot of read-modify-write operations taking place constantly. Bandwidth utilization is also an important parameter. With general IT data, there are huge peaks and troughs in bandwidth usage due to fluctuating demands.
On the other hand, video surveillance data has entirely different characteristics. All the cameras generate data in parallel, at a variable but consistently high rate and all throughout the day. This rate is even higher when the recorded scenes are “busy” or poorly lit like night scenes which generates more image noise. In addition, surveillance video data is almost unidirectional (i.e., into the storage system), with 100 percent of the data being recorded and only a small fraction, about 1-3 percent ever being retrieved for review and playback. Another significant difference is that video data are never changed, only overwritten; they have a precisely defined lifetime before deletion.
“Understanding these differences allows the design of a system which significantly reduces the three key causes of hard disk failure: temperature, vibration and wear,”explained McLeod. For example since the writing to the disks is sequential, the writing head can move slowly across the drive, reducing temperature and wear. In addition, non-writing disks can be switched off saving further on wear and reducing power consumption.
Current Trends Will Continue
The trends of the past few years will likely continue in the near future. Cameras will continue to deliver more information and higher resolutions and the need for storage will increase. Even if codec technology manages to significantly reduce the storage requirement per channel, users are more likely to install more cameras or record higher quality video. The vast amounts of bandwidth needed to upload video data to cloud-based data centers are still not available and the industry will have to continue to rely on local storage. Design and selection of the storage solution will probably only increase in importance.