Image via Wikipedia

Distributed video storage solution = $900/month.

Centralized storage solution (180T storage using NAS storage solution offsite at a data center)= over $228,000 per month.

‘Nuff said?

I was inspired to write this from John Honovich’s blog post entitled, “Why Centralized NVR Recording Does Not Work”. I wanted to take it a step further with a financial comparison.

If you look at the costs, there’s no comparison. For example, take a business which has 1,500 remote locations, each with 5 cameras and a requirement for 30 days of storage per camera. This requires on the order of 180 terabytes of video storage.

Distributed solution (storage at the remote sites) = $900/month

• That’s roughly $32,400 in capital for the storage at all locations which has an average life of 3 years.
• All storage is local to the site, video is only streamed on demand when viewed by a user using the existing retail class broadband circuit at the remote site.
• 500GB drive at $0.18 per GB
• If you’re not willing to take the risk of a disk failure at an individual site, losing video for that one location, then double the cost for redundant disk drives

Centralized storage solution (180T storage using NAS storage solution offsite at a data center)= over $228,000 per month.

• Storage: $180,000 in capital or $3,750 per month just for the storage equipment. Centralized storage requires network attached storage (NAS) and due to aggregation requires at least RAID5 for redundancy due to risk of losing large quantities of video across multiple locations.
• NAS RAID5 storage at $1 per GB
• Network capacity: Add in the network capacity required for offsite storage, and you’re looking at an extra $300,000 per month for single T1 circuits to each of the 1,500 locations. (I’m guessing an average of $200/month for a single T1 to the Internet)
• This is excluding costs for data center facilities (cooling and power) and also excludes network bandwidth charges for the 1.5 to 2 Gigabits per second of network capacity required at the data center.

It’s likely that single T1 circuits would not actually provide enough capacity for 4 cameras, but I’ll give the benefit of the doubt due to use of advanced h.264 video codecs and excellent motion detection to reduce the bandwidth demands.

The bandwidth to the remote locations obviously drives the vast majority of the costs. This bandwidth is required when streaming all video back to a remote location.

So instead of buying all that network capacity, distributed storage keeps all the video on site and only sends video over the network when needed. This could be when someone views the video and could include the occasional video upload as well (perhaps automated based on special events). 99% of the recorded video is never viewed and never transmitted.

There’s no reason to send all that video back to a central location when the vast majority is never used.

The only real challenge for the distributed storage is scaling the management of all the remote devices. That’s where MVaaS comes in. When one has a lot of remote storage devices which have moving parts (ie: hard drives) one must have automated systems to monitor and report failures. You’re going to need a video management system to enable viewing your 6,000 remote cameras anyways, so the system should also support managing the distributed storage for those cameras.