How to design a video surveillance solution

Seven fundamental questions you should be asking as you design for surveillance


While storage is always getting cheaper, video surveillance demands huge amount of storage. For comparison, Google's email service offer about 7 GB/s of free email storage. This is considered to be an enormous amount for email. However, a single camera could consume that much storage in a day.  It is fairly common for video surveillance systems to require multiple TBs of storage even with only a few dozen cameras. Because storage is such a significant cost, numerous techniques exist to optimize the use of storage.

The type of security threats also impact determining storage duration. For instance, a major threat at banks is the report of fraudulent investigations. These incidents are often not reported by affected customers until 60 or 90 days after the incident. As such, banks have great need for longer term storage.  By contrast, casinos usually know about issues right away and if a problem is to arise they learn about it in the same week.  Casinos then, very frequently, use much shorter storage duration (a few weeks is common).

Three fundamental types of storage may be selected:

  • Internal storage uses hard drives built inside of a DVR, NVR or server.  Today this is still the most common form of storage. With hard drives of up to 1 TB common today, internal storage can provide total storage of 2TB to 4TB.  Internal storage is the cheapest option but tends to be less reliable and scalable than the other options. Nonetheless, it is used the most frequently in video surveillance.

  • Directly Attached storage is when hard drives are located outside of the DVR, NVR or server.  Storage appliances such as NAS or SANs are used to manage hard drives. This usually provides greater scalability, flexibility and redundancy. However, the cost per TB is usually more than internal storage. Attached storage is most often used in large camera count applications.

  • Storage Clusters are IP based 'pools' of storage specialized in storing video from large numbers of cameras. Multiple DVRs, NVRs or servers can stream video to these storage clusters.  They provide efficient, flexible and scalable storage for very large camera counts.  Storage clusters are the most important emerging trend in video surveillance storage. Learn more about storage clusters for video surveillance.

5. Video Analytics

Video analytics scan incoming video feeds to (1) optimize storage or (2) to identify threatening/interesting events.

Storage optimization is the most commonly used application of video analytics. In its simplest form, video analytics examines video feeds to identify changes in motion.  Based on the presence or absence of motion, the video management system can decide not to store video or store video at a lower frame rate or resolution. Because surveillance video captures long periods of inactivity (like hallways and staircases, buildings when they are closed, etc.), using motion analytics can reduce storage consumption by 60% - 80% relative to continuously recording.

Using video analytics to identify threatening/interesting events is the more 'exciting' form of video analytics. Indeed, generally when industry people talk of video analytics, this is their intended reference. Common examples of this are perimeter violation, abandonded object, people counting and license plate recognition. The goal of these types of video analytics is to pro-actively identify security incidents and to stop them in progress (e.g., perimeter violation spots a thief jumping your fence so that you can stop them in real time, license plate recognition identifies a vehicle belonging to a wanted criminal so you can apprehend him).

These video analytics have been generally viewed as a dissapointment. While many observers believe that video analytics will improve, the video analytics market is currently contracting (in response to its issues and the recession). Learn more about the challenges in video analytics.

6. Viewing Video

Surveillance video is ultimately viewed by human beings.  Most surveillance video is never viewed. Of the video that is viewed, the most common use is for historical investigations. Some surveillance video is viewed live continuously, generally in retail (to spot shoplifters) and in public surveillance (to identify criminal threats. Most live video surveillance is done periodically in response to a 'called-in' threat or to check up on the status of a remote facility.

4 fundamental options exist for viewing video.