1. Skip to navigation
  2. Skip to content
  3. Skip to sidebar

Recording at the edge – A new approach to surveillance system design

Edge recording is a term which is has started to populate through the IP camera industry with regards to network video recording. Edge recording is a network configuration designed to reduce the bottlenecks inherent to centralized video systems. This article describes how edge recording functions work and describes the pros and cons of its use.

Conventional Security System design

With conventional recording applications there is a centralized recording server or Network Video Recorder (NVR). The centralized server pulls a video feed from each camera continuously which is then recorded into storage. If the recorder is set to record from motion detection a stream is still continuously pulled from each camera. The system then looks at each of the video feeds to determine if there is any motion. If motion is detected the video is recorded.

This design comes from older analog systems which used a coaxial cable to transmit video but for network video applications things can be done a bit more intelligently.

The truth about centralized recording

There are a few common problems with the centralized approach to network video.

Diagram showing 5 cameras conntected to a centralised recorder


Firstly, this centralized approach is very bad for bandwidth. A stream is pulled from each camera continually, regardless of whether recordings are continuous or based on motion detection. The more cameras that are added the worse the problem gets until your network is completely saturated.

Server Specification

With so much data being analyzed simultaneously, a centralized recording server needs to be capable of processing all the footage and must have enough bandwidth to receive all video streams at the same time. This makes the hardware more expensive and can often lead to compromises being made on image quality to reduce hardware costs, reducing the effectiveness of the system.

Fault tolerance

Because all footage is being recorded to a centralized server, if the server goes offline for any reason, no video will be recorded. Furthermore, if the server is damaged, all recordings are lost.

A new approach

Edge recording (otherwise known as decentralized design) is a method which removes the dependency on the centralized server. With the older approach, cameras were essentially dumb units where the server performed all of the processing and held all of the recorded video footage. The new approach moves the recording functionality away from the server and onto the camera itself.

Depending on the camera specification, footage can either be stored to local network hard drives or directly to the camera using flash media or internal hard drives.

Diagram of cameras attached to network drives which connect to each other

Fringe Benefits

With all recording being performed on the camera and stored either to network attached storage or directly onto the camera, all recordings are transferred locally and do not have an impact on the rest of the network.

Motion detection recording is also greatly optimized. With the motion detection settings being controlled by the camera, no footage is sent from the camera until a motion event is detected. Once detected, footage will be recorded and will stop again as soon as the motion ends.

Because all of the processing and recording functionality has been transferred from the storage server to the camera, the storage drives are not as sophisticated and as a result are a lot cheaper to manufacture. The storage servers are essentially only hard drives which are accessible from the network.

The reliability of the system has also been increased. If any of the network storage drives go offline, only the cameras recording to that server will be affected. All other cameras remain online.


With storage being distributed throughout the network, management of storage locations and camera footage can be more difficult. Some camera manufacturers have tried to combat this. Mobotix for example feature MxControlCentre which can let you monitor footage from multiple locations throughout the network.

Because of the additional processing power required by each camera, the cost of each device tends to be more than a similar camera without the recording functionality. Also, if you are using network-based storage and wish to distribute the footage across the network, you need to have more than one device, increasing costs.

ONVIF also poses a potential threat to edge-based systems. ONVIF (Open Network Video Interchange Forum) is a partnership between the main manufacturers in the network video industry. The partnership is designing a shared protocol for exchanging video data between devices from different manufacturers. Despite the fact that the protocol is still in its infancy, it is based on a centralized design. If the protocol is finalized without any support for edge-based network designs, fewer and fewer cameras will support edge recording as the popularity of ONVIF increases.

The best of both worlds

There are cameras which try to combine the benefits of both edge and centralized systems. The Panasonic i-Pro range, for example, stores footage to a centralized server. Should there be any network downtime, or problems with the centralized server, images are recorded on SD cards in the camera. Once network connection is restored, footage is transferred in the background so no footage is lost.

This approach improves the reliability of the system but still requires a highly-powerful centralized server to handle footage, causing data bottlenecks and increasing costs.


Both sides of the argument have valid reasons why you would choose one over the other, namely the reduction in bandwidth and fault tolerance of edge recording and the easy management and lowered per-camera costs of centralized recording.

However, we feel that the older centralized design has had its day. Just as analog systems are being upgraded to IP, system design should also be updated accordingly.

Edge recording should be the way that professional CCTV security installations are designed to reduce bandwidth saturation and increase fault tolerance. However, we also feel that the additional costs, both in camera and storage devices, coupled with the industrial momentum, currently being reinforced by the preliminary ONVIF design structure, may mean that the industry will continue to adopt a centralized design for the foreseeable future.

Published on February 26th, 2010 by James Drinkwater

Comments are closed.