Pivot3 Blog

DAS Behaving Badly - Part 2

Poor Fault Tolerance 

In our last post on DAS behaving badly, you learned that video surveillance systems write data to disk 24x7x365 and 4-5x more often than typical IT applications. This write-intensive application puts significant strain on NVRs’ storage components, making them 3x more likely to fail than when used for standard IT applications.

Furthermore, the amount of data being generated by IP video systems is doubling roughly every 18 months.

NVRs are simply generic servers with built-in (“directly attached”) storage, designed for standard IT workloads. When used for video surveillance this architecture is highly prone to failure, leading to the permanent loss of recorded video and significant performance degradation that prevents video from being properly captured and stored.

To guard against this, most modern NVRs use a technology called RAID, developed in 1977 to protect data from hard drive failures. Nearly 40 years later, RAID is still by far the most prevalent means of fault tolerance, but can only protect against one or two drives failing simultaneously.

Additionally, as drive capacities reach 8 and 10TB, rebuild times can last days or even weeks and put major strain on system resources, significantly slowing performance and increasing the likelihood of another failure that will lead to permanent data loss.

Most IT professionals believe RAID is entirely impractical for modern technology. And for surveillance systems that cannot tolerate ANY downtime or data loss, NVRs are more susceptible than ever before.