How to Optimize NVR / DVR Storage

Megapixel cameras have brought renewed interest in measures to maximize NVR/DVR storage duration and use. Cost is a big factor as the potential storage needed could increase by 10x or more than historical standards. Understanding what options and measures are available is becoming increasingly important to selecting NVRs/DVRs and designing IP video systems. This report surveys common measures used to maximize strorage duration and use.

Recently, interest has risen in new product categories that specialize in optimizing storage use.

Frost and Sullivan has recently reviewed "Video Lifecycle Management Solutions" and identified TimeSight Systems as a "young leader" in this space. The release does a good job of identifying the problem and highlighting one potential solution.

As almost all video systems have numerous measures to optimize storage use, I recommend that integrators and end users focus on utilizing existing measures in leading systems. Video system developers have been building tools for years to address storage optimization. Most will be best served by selecting a video management system based on features optimized for your specific security needs. Significant and comparable storage optimization can generally be accomplished on most mainstream NVR / DVR systems.

How Do I Optimize Storage?

This report reviews 8 commonly available storage optimization functions available on mainstream NVR /DVR systems. Though not every system has all of these features, all systems offer a number of them, providing strong storage optimization.

Here is the list:

• Basic Motion Analytics
• Advanced Video Analytics
• Motion Exclusion Zones
• Data Aging
• Recording Schedule
• CODEC Selection
• Dual Streaming
• Storage Clusters

Advanced Video Analytics

Now that video analytics are getting accurate at detecting people, faces and vehicles, this intelligence can be used to control recording. I believe this will become one of the most powerful new areas of storage optimization in the next 3 years. Long term storage can be optimized by selectively recording objects most likely to be of long term interest - people, faces and vehicles. Traditionally, long term storage optimization techniques reduce the quality or the frame rate of all video uniformly. With video analytics, storage optimization techniques can become smarter, increasing the probability of possessing quality long term evidence while minimizing total storage consumed.

For instance, in addition to recording video, 3VR records all faces seen on cameras. Faces of all the people (100,000+) conducting transactions at a bank branch can be stored at 4CIF quality with less than 20GB of storage. This is 1/100th the amount of storage needed for video and the most important evidence for retail bank's security needs. Of course, today this is just faces but the same process can and will certainly eventually be used to store all the people seen, all the cars moving through an area, etc.

Video analytic companies specializing in perimeter violation are reducing storage needs for those cameras by 90% or more. By placing intelligence in the camera, the camera can only stream or the management system can only record specific objects of interest. For cameras whose main purpose is real time alerting, this is a great storage win. Of course, many cameras are needed for investigation purposes and need storage. As such, this is simply another tool in our collection.

Basic Motion Analytics

Most video surveillance deployments use basic motion analytics to control recording. Because most facilities have significant periods of low activity (e.g., nights, weekends) and areas of low activity (e.g., hallways, stairwells), motion analytics can reduce storage consumption by 50% to 80%. Most systems set their basic motion analytics to be fairly conservative so that they rarely miss real incidents. As such, basic motion analytics is trusted and used by many military bases, banks and Fortune 100 companies and most real world deployments. Of course, some facilities do not want to take any risk and require continuous recording.

A nice balance that is sometimes achieved is a combination of continuous and motion based recording with a baseline level of continuous recording (e.g., 3 frames per second) and motion based recording set higher (say to 15 fps). This ensures that video is always recorded but storage use is optimized for when activity of interest is most likely to occur (that is, when motion is detected).

Motion Exclusion Zones

Using basic motion analytics to control recording is enhanced through using motion exclusion zones. It is common for cameras to cover areas that are not of interest to users. Examples include highways behind the building, a tree out front, windows, ceiling lights, etc. Taking a few minutes to set up motion exclusion zones can reduce the storage utilization by up to 50% on certain cameras. After the first week of a new install, a review should be conducted to tune these settings.

Data Aging

Many systems reduce the number of frames in stored video as the video is older. The basic premise is the older the video, the lower the probability that the video is relevant. Rather than simply delete the video, the size of the video is reduced so that some evidence is available just in case but the storage costs are minimized.

For instance, March Networks has a feature called "Intelligent Video Retention." Avigilon has an advanced data aging solution that specializes in optimizing storage for multi-megapixel cameras. In higher end video systems, this type of feature is frequently available. It's quite useful because it can easily double storage duration.

Recording on a Schedule

Many organizations have greater security risks at different times of the day. Schedules are a common feature to adjust recording parameters to match those different level of risks. For instance, an organization may want continuous recording during business hours but is ok with only having motion based recording after hours. Making this adjustment can reduce video storage use by up to 40%.

CODEC Choice

Choosing a video CODEC that provides the most efficient storage utilization has been a key component of video system designs for years. While technical issues exists, the trend of moving from less efficient to more efficient CODECs is clear (e.g., from MJPEG to MPEG-4 to H.264). The key practical issue currently is the use of H.264 for megapixel cameras due to the high system requirements H.264 demands. With multiple megapixel manufacturers releasing H.264 megapixel cameras, in the next few years H.264 megapixel cameras looks certain to be a reality (at least for lower resolution MP cameras). Migrating from MJPEG to H.264 can reduce storage use by 50% or more.

Dual Streaming

To maximize CODECs different strengths and weaknesses, multiple video streams can be used. For instance, H.264 may be the best choice for storage optimization but MJPEG has advantages of live monitoring (e.g., lower delay, lower processing power to view). Most cameras support dual streaming. Video surveillance systems can take advantage of this to reduce storage costs while ensuring optimal live video monitoring.

Storage Clusters

Historically, storage was separated into small pools for each unit and options for upgrading storage were limited to a few TBs (at most). Today, with storage clusters for video surveillance maturing, centralized pools of storage can create ~ 30% efficiencies in storage use and make extending storage simple and fairly inexpensive. If you are interested in learning more, read my extensive analysis of storage clusters.

Conclusion

While by no means comprehensive, this survey should help engineers and users to identify and use commonly available measures to optimize storage duration. Understand what your systems offer and make use of those features. By doing so, you will be able to optimize the storage of most any DVR or NVR and accommodate increasing storage demands.