Eric Fullerton, Chief Sales & Marketing Officer, and Christian Bohn, VP Marketing & Product Management, have an open chat about the current status, advantages and limitations of using storage at the edge for IP video surveillance data.
Fullerton: At the Milestone Integration Platform Symposium (MIPS) partner event in February of this year, we demonstrated with Sony the ability to take the network connector out of the camera, put it in front of the camera so you could see it was not connected, put it back in the camera, then go into the Milestone recording software and see us holding the plug in front of the camera. So you can take the network connection away from the camera, yet our software makes sure that whatever was recorded while the camera was unplugged, recording only at the edge, also gets transported afterwards back into the Milestone recording server for the archives. This edge storage will be supported in the coming release of XProtect Corporate 4.0.
Technologically, the SSD card (solid state drive) brings the edge storage promise that the camera can become like a DVR, self-containing, to record and store on its own. This makes the system more robust because the network is not necessary until archiving is required. However, the camera can only hold so much: it fills up with the data and starts deleting as new video comes in over it, so you still need to send it off to a recording server.
The challenge then is getting the individually edge-stored video data to fit into the relevant time sequence in the recorded archives after the fact. Milestone's software would then be treating the data from that camera as it would from a DVR or other recorder.
Bohn: At this time, the edge technology isn't developed enough with sufficient size for on-board camera storage to host the data for very long. This will change in a few years, of course. You can get SD cards today that are around 160GB and if you use H.264 with good frame compression, you can save a lot but not enough for a week or more.
Fullerton: With edge storage making the camera become its own mini recording server, the storage on the camera works as a buffer when you're not connected to the network, and the video can be transported into the main database later. You can use it to only download the parts that are of interest which could help make surveillance run without high cost wide-area bandwidth usage.
Bohn: The initial scenario for using edge storage is for a few minutes or hours when doing scheduled system maintenance and switch connection is lost to the cameras. You could trigger the cameras to preserve data offline at the edge, keeping the data for evidence so nothing is lost during that time, and then download it afterwards to the central server for the full chain of evidence.
Fullerton: Analytics, motion detection, time periods or other rules in our recording software can determine which video is necessary to be kept. For example, for use in buses it's a natural approach because the driver would know if anything happened that needed to be downloaded back at the central station or wherever the main recording database is located. This technology is also interesting wherever there is intermittent network capability: poor infrastructure, places with harsh weather environments that create power outages, etc.
Bohn: Camera manufacturers need to do anything they can to differentiate themselves. They need to stay proactive for the market, so edge storage is a good idea for them. Other scenarios for its use include when a network switch dies, or a system administrator does something in the configuration or during maintenance work, when you would have the option of slotting the video recording into the edge storage at the camera. The key parameter here is no loss of data: increasing the up-time of your recording.
However, you would not be able to see the video live - you would be in a reactive mode since you could only see the data after the fact. Storage at the edge is for evidence collection and investigations - you still need a central recording database for archiving. Without a central server, you have to rely on the performance of the cameras alone. If they go down – and cameras do – you lose that surveillance capability. Servers and redundant servers are much more robust than cameras, as hardware goes. Large and critical security application sites would therefore never go to a pure edge-based model.
Fullerton: There's discussion in the market about the possibility of server-less surveillance where all the surveillance will run only on the cameras, viewed in the cloud via a browser. We think that could be too unstable. It would only take one or two instances of users not being able to see all their cameras before they'd drop that approach.
Bohn: In summary, the key reasons for storage at the edge are reliability: availability of data up-time for the entire surveillance installation and the need to ensure continued evidence collection while doing system maintenance (a normal IT procedure for any business), power outage environments from harsh weather or poor infrastructure, and wireless or mobile transportation installations with intermittent network connections .