How Much Storage is Needed for Video Surveillance?
How much storage is needed for video surveillance is an important question for planning new system deployments and for determining total installation cost.
In the last few years, fears have increased dramatically that the amount of storage needed was exploding and that video surveillance systems would need 10 times or 100 times the amount of storage they historically used. Coming from a world where 250GB or 500GB DVRs was common, this created a lot of anxiety and questions.
Rules of Thumb
The following two metrics should provide a range:
- 80% of surveillance cameras use between 60GB to 600GB of storage
- 99% of surveillance cameras use between 6GB and 6TB of storage
Such broad ranges may seem surprising but there are a dramatic number of factors that influence video surveillance storage (see our calcuating storage report).
In this report, we examine a number of common scenarios to help you better understand whether you are likely to need 6GB, 60GB, 600GB or 6TB.
Despite the complexity, the good news, though, is that fear of explosive storage growth are unfounded. While more storage will be used per camera in the future, it will match the natural decrease in storage prices that happens continuously.
In this report, we examine how changes in storage pricing, cameras deployed, CODECs used and evidence quality required will impact storage utilization.
The following are rules of thumbs that are meant to help in estimation. As we examined in the calculation report, there are so many factors and many of the key ones are difficult to project, these are guidelines to help find the right range.
6 may seem an odd number but when multiplying it with the most common number of cameras used in video surveillance - 8, 16 and 32 - the results are 'even' numbers. For instance, if 60 GB comes closet to your scenario, 60GB x 16 cameras is approximately 1 TB. Alternatively, if 600 GB is a better match and you have 8 cameras, then the approximate storage needed is 5 TB.
Also, by choosing levels that are a factor of 10 from each other, you will quickly be able to determine if SSD, NAS or internal storage are feasible.
6 GB - Low Resolution/Motion Only/Moderate Duration
At about 6 GB of storage per camera, your options are very limited. If you record continuously, you can only stream 80kb/s for 1 week. 80kb/s is quite low - barely enough for most CODECs for CIF/5 fps.
The only reasonable way to use 6GB for 1 month is to use motion based recording for scenes with low to moderate activity.
6GB is pretty much impossible for megapixel recording at any length. Even an H.264 megapixel camera with moderate motion during the day will consume 6GB in less than 7 hours.
If 6GB (or slightly more) is sufficient for your needs, it is possible to use SD cards for storing video on-board the camera. However, retrieving the video could be an issue with many VMS.
For storing in DVRs/NVRs/servers, one can store way more than 6GB per camera without any significant increase in cost. For instance, 16 cameras at 6GB each is only 100 GB of storage (which is very low given today's hard drives). Most users are better off using higher resolution or storing for longer durations than the constraints forced by 6GB.
60 GB - SD Resolution/Frame Rate and Moderate Duration
60 GB is likely the most common storage consumption in today's video surveillance systems. Unlike 6GB, it's large enough to record quality video for some period of time but it is not too expensive. For instance, 16 cameras consuming 60GB storage each is 1 TB - which is the sweet spot of today's hard drives.
With 60GB, motion based recording is possible for multiple months at up to 4CIF and less than 10 fps. Continuous based recording for full frame rate/SD resolution at 1-2 weeks is possible.
With somewhat more bandwidth (say 120 GB to 240GB), almost all Standard Definition (non-megapixel cameras) can be recorded continuously and stored for at least 1 month, if not longer.
At this level, cameras can be stored using internal storage on DVRs/NVRs/servers without any significant problem (1TB to 2TB for 16 cameras is sufficient).
600 GB - H.264 Megapixel/Moderate Duration
The next big step up is storage consumption generally comes from using H.264 megapixel cameras.
While a standard definition H.264 camera may need 500 kb/s to 1000 Kb/s, a 1920 x 1080 megapixel routinely needs 4 - 8 times the amount of storage.
For instance, 1 month of continuously recording a 1080p camera streaming at 2Mb/s is about 600 GB. 2 Mb/s is at the low end of H.264 megapixel consumption - higher motion and lower light conditions can double or triple consumption.
At the same time, a 1080p camera using motion based recording should be able to record multiple months with 600 GB of storage.
If you need 600 GB per camera, you are pushing the upper limits of what DVRs/NVRs/servers can handle (16 of such cameras would require 10 TB). At this consumption level, storage clusters are likely the most economic option.
6 TB - MJPEG Megapixel/Moderate Duration or H.264 Megapixel/Long Duration
The real 'scary' storage levels typically only because of using MJPEG megapixel cameras. For instance, if you were using SD cameras, it would require 2 years of continuous recording to hit 6TB. However, with MJPEG megapixel cameras, 6TB can be consumed in little over 1 month.
Such situations result in anecdotes about users spending $9,000 USD in storage per camera.
At 6 TBs per camera, storage clusters are really the only practical option. At this level you mainly need to decide about whether to use a video surveillance specific cluster or a general purpose SAN.
Projections on Future Storage Use
It's the last scenario that has triggered alarms in the video surveillance business. If cameras now require 6Tb of storage each, it's going to cause massive issues with DVRs, traditional servers, budgeting, cameras selection, etc.
However, I believe the peak of storage fears passed in 2007-2008. There's a number of off-setting technological advances that have begun and will continue to occur that will eliminate any potential 'storage crisis.'
The most obvious, but least important, is that HD storage capacities continue to increase. Where 2 years ago, 500GB - 750GB were the high end, today it's double that -- 1-1.5TB.
Looking 5 years, it's reasonable to expect storage capacity to quadruple (i.e. doubling twice). In 2014, a 4GB - 6TB hard drive for about $100 USD looks realistic.
Between 2006 and 2009, the big jump in video surveillance resolution occurred. We are unlikely to see such an important jump for years to come (if ever). Moving from SD (.3 megapixel) to 3MP/1080p makes a huge practical difference. It's an almost 10x increase in pixels and, more importantly a 300% increase in pixel width. Moving from 3MP to 5MP is almost the same in terms of number of pixels but far less in terms of practical impact. As such, it's likely that cameras over 3MP remain niche for years to come.
Furthermore, 1080p resolution is good enough to satisfy many indoor cameras. A major problem with traditional cameras was that SD resolution could only provide facial details at 10 feet (3 meters) or less. There are lots of areas where this is simply too narrow (forcing users to give up details to get the wider area). With 1080p, a camera can now provide facial detail at up to 30 feet (10 meters). This is good enough for many more applications (because physical barriers, like walls, tend to limit how wide an area a camera can cover). Because of this demand for even more resolution should be lower.
Finally, it's not likely that video surveillance manufacturers will even offer cameras in the mainstream market beyond 10 or 12 MP, even 5 years from now. 5MP is the mainstream maximum now and there's little imminent push to move beyond this.
Many manufacturers trumpet how video surveillance users are demanding longer storage duration and higher quality. While it's absolutely true that they are using higher quality and storing video longer, this is mostly driven by falling storage (and megapixel camera) costs. We examined this issues previously in our analysis of video surveillance storage demand.
Part of the fear in 2007-2008 was based on the assumption of using MJPEG for megapixel. If MJPEG was not superseded, this would have caused a problem. However, it's clear now in 2009, that H.264 will replace MJPEG as the standard CODEC for megapixel cameras (most of the MPEG megapixel vendors I talk to who have not publicly announced plans are working on releasing H.264 models in the next year).
While H.264 megapixel is not meeting the marketing hype of 90% bandwidth reduction (low light and high motion being too main problems), it is still substantially reducing the storage load relative to MJPEG.
Even more importantly, by 2011/2012, expect many vendors to be releasing H.264 SVC codec support. SVC will have a big impact on total storage consumption as it allows for reducing stream size over time (by reducing resolution and/or frame rates). This could drop total storage consumption by 50% or more. For instance, a camera could be set to record at 30fps/1080 for the first week, then drop to 5fps/1080 for the remainder of the month and then drop again to 1fps/4CIF for the next 2 months. This is just an example but the total storage needed for a 3 month period would be dramatically lower than keeping the video at 30fps/1080 for the full 3 months.
Today, 60 - 240 GB per camera is a common storage level. With more megapixel being used and storage capacities continuing to increase, in the next 5 years, it is likely that 480 - 600 GB per camera becomes the new 'norm'. However, this should remain of similar cost/affordability.