There is arguably no higher cost component in a video surveillance system than video storage equipment. Coupled with the proliferation of CCTV cameras, the ability of network-based video systems to enable easy viewing of live and stored video at multiple locations has driven video storage to higher levels of importance and utility. There is no doubt in my mind that not only will the number of stored bytes of video data continue to rapidly increase, but also that offsetting advancements in a number of areas will keep costs reasonable.
Start with the cost of the storage media itself. In 1950, the cost per megabyte of storage was approximately $10,000. In the early 1980s, that cost had fallen to less than $500; by 1990, under $10; and by the late 1990s, about a penny. In 2001, we saw the cost per gigabyte less than $5; under $1 in 2004, and less than a dime in 2010. Today, you can purchase a 3 TB SATA drive from Amazon for $159. Granted, these are only drive costs and do not reflect the hardware and software around them to make them work, but the downward trend in raw storage costs continues. This will only fuel the appetite to store more.
However, having more storage does not come without a price. The concentration of servers and storage has both a facility footprint and energy impact. Several different technology approaches promise to address this issue, and, hopefully, ease the load. These include storage in the cloud, storage on the edge and storage that does not involve spinning drives.
Storage in the cloud — which I use myself — has come of age. While this topic is the subject for another article, it is worth mentioning that centralized cloud-based storage can provide enhanced levels of redundancy and security (if properly implemented), while offering the opportunity for new services such as analytic-based search.
What about those cases where you really do not need to store all the video that is generated by every camera? For this, there is storage on the edge, using flash, SD, or CF, for example, capable of storing hours of video and streaming in conjunction with an alarm event, including pre- and post-alarm. An alarm is defined as any event that is determined to be of interest according to the rules that are set up. No longer relying on external inputs, cameras are smart enough to determine what is a problem and what is not. They can send notification — perhaps including a clip in an e-mail — and the video upon request.
Also emerging is solid state drive (SSD) technology. Currently, about $2 per GB, Moore’s law would predict a roughly 50-percent cost drop per year. Still significantly higher priced than magnetic SAS or SATA drives, SSD technology offers higher performance (approximately 200 times the read speed) and great potential in an edge device. Somewhat surprisingly, a quick check of the power consumption of 1 TB SSD and SATA drives showed no appreciable difference. I found a 1 TB Hitachi SATA drive which is specified for 5.2 W at idle vs. 4.3W for an OCS SSD SATA drive; however, MTBF on the SSD drive was 2,000,000-plus hours. So, if the read performance is not required for the video, the cost differential will likely constrain SSD drives to the edge and other specialty circumstances.
Other important developments are occurring in the software that manages and searches stored video. Storage appliance vendors have made great strides in their software/OS to optimize video storage and retrieval. Managing the block transfers of video data and minimizing fragmentation improves the overall data reliability of the stored video. Spinning down the discs for archived video until required can result in significant energy savings — although this gets a little tricky with RAID drives, where information is written across multiple drives to allow the data to be reconstituted in case of a drive failure.
True operating costs are reflected not just in facility and energy, but also in the man-hours it takes to search the stored video. Several suppliers have evolved techniques to make the search process far more efficient. One promising technique I have witnessed involves distilling an hour of video into two minutes using a technique called “Video Synopsis.” Reminiscent of fast forwarding through video tape, the displayed video appears at times ghost-like with people appearing to walk through people. The net effect, though, provides a true sense of what occurred during that time period.
Another approach uses intelligent search, where rules based on analytics or other metrics can pick out those pieces of video that are relevant. For example, find all of the clips of the person in the red sweater and dark pants in the mall from 2-3 p.m.
Other human factors play a major role in the true costs of storage, starting with installation and training. Industry vendors are improving the means by which they deliver both training and the ease of set up and use of their equipment. This task has been made simpler by the increasing competence of many integrators in our industry who have stepped up efforts to upgrade the quality, knowledge and competence of their technical staffs. Some have reached the point where they now purchase and configure conventional off-the shelf storage, reporting good results and nice margins.
Undoubtedly, they would draw an argument from industry vendors who have optimized their gear for video and claim superior performance and reliability. This is one dispute I expect to continue for a while — that’s good though…it shows progress.
