Is the Surveillance Industry Ripe for a Storage Change?

Details

Created: Friday, 21 July 2017 16:05

Hits: 3551

Picture this scenario: During the nighttime, there has been a break-in at a company location and the perpetrator has been caught on tape. The recording is stored and reviewed at a later time, only to find that -- due to compression -- the images of the perpetrator is obscured; also, frames have been dropped because of unstable data recording. This makes identification much harder and the recording might even be inadmissible in court.

Now let us tweak this scenario a bit: During the nighttime, a person is picked up by the surveillance cameras. With fast-working data analytics, the system notices that there is someone in the restricted area and simultaneously notifies the appropriate personnel. The infiltrator is quickly apprehended on the premises, and the high quality of the recording helps correctly identify the person. So, what can this example tell us?

The thing is that most surveillance systems today use hard disk drives (HDDs) for storage purposes. While this works perfectly well for basic setups, it can be a hindrance for implementing modern surveillance features. The use of solid state drives (SSDs) for data recording is a solid step forward (pun intended) towards fixing this, as it gives the user access to data analytics that enables advanced features such as facial recognition and automated traffic control systems.

This raises the question: "What is the reason HDDs are still so commonplace?" In fact, this is partially due to some misconceptions that still linger, such as, "HDDs are more cost efficient and have higher capacities, while SSDs have relatively low capacities and are more prone to failure due to P/E (program-erase) cycle limitations." As technology advances, however, these "truths" are not nearly as clear-cut anymore. Furthermore, as the relative price gap between HDDS and SSD is closing, SSDs can now provide a cost-efficient alternative to many traditional surveillance setups.

Puncturing the low endurance myth
Let's start off with limited P/E cycles. This has been seen as the bane of any SSD in a recording-heavy environment, and -- 10 years ago -- this by itself would be more than a good enough reason to look elsewhere for a storage solution. But things change, as the slightly informal -- yet very extensive -- SSD test from The Tech Report clearly shows.

The SSDs involved all went through an absurd amount of around-the-clock write and erase cycles, yet -- even with some hiccups -- they all performed well past their stated limits. By also taking into account advances in technology since the test was started (2013), coupled with the fact that these were all consumer-grade devices, we can safely put the final nail in the myth of limited SSD endurance. With the correct setup, there's simply no need to worry that an endurance failure will ever happen.

Now to be fair, running out of P/E cycles is only one way that an SSD can fail; it can still struggle with faulty ICs, bad batches and so on. So, some SSDs will still fail, never having gotten close to their limits. If we look at the data for SSD and HDD reliability, while SSDs will experience smaller failures, the rate of complete failure is much higher for HDDs. Although this could be seen as comparing apples to oranges, there is certainly little ground to claim that the HDD has any advantage over SSD when it comes to endurance.

Quality and analytics
Another even more clear advantage for the SSD is speed, which has also been the main selling point ever since this form of storage entered the market. Modern surveillance setups don't just record and store data, but they are also seeing increasing requirements for data analytics. This means handling on-the-spot operations like facial recognition and then taking immediate action once a hit is made. Thus, the system has to be able to handle demanding simultaneous read/write operations, which HDD technology is -- in many cases -- poorly equipped to handle.

Real-time analytics of this sort is also reliant on high quality recording, and is susceptible to frame loss if the system is unable to sustain stable recording rates. SSDs can, through firmware optimization, more easily fill this role than any HDD, as they are much more adapt at both random and simultaneous read/write operations. The graphs below show just how large an effect optimization can have on video recording stability.

Write performance for standard SSD without firmware optimization (Source: Innodisk)

Write performance for Innodisk SSD with firmware optimization implemented (Source: Innodisk)

Capacity and price
But even with the two elements discussed above working in favor of the SSD, there remain two bastions where the HDD still stands strong: namely price and capacity. For those just looking for a simple and affordable system, an HDD-based solution can be a perfectly good fit.

This holds true for low-performance HDDs, but once you reach the 10K-15K RPM drives, it's a different story. These drives have platters that spin up to twice as fast as normal drives and have significantly increased performance. However, prices increase and capacities drop dramatically, even being surpassed by higher-end SSDs.

So, if your application needs to perform above that of a basic surveillance setup, instead of moving up the HDD ladder, the easier solution is to just skip this rung altogether and instead opt for an SSD. By also calculating in a higher overall reliability, the total cost of ownership will -- in many cases -- tip the scales in favor of the SSD solution.

Innodisk's 2.5" SATA SSD 3MV2-P offers up to 2TB capacity coupled with
high speed and stable performance (Source: Innodisk)

A technological impasse
Another aspect to consider is the underlying technology. Although referred to as drives, SSDs don't actually have any moving parts. This means that they are naturally more adept at taking a beating both in terms of shock and temperature; it also means less heat created during operation. By comparison, an HDD has a head that flies on a thin layer of air above the platter. Any strong enough vibration or shock will lead to a head crash and data being lost. Thus, for any recording performed onboard vehicles or in hostile environments, an SSD should be the primary choice for storage.

HDD components are also more limited by physics itself; you can only go so small before losing mechanical integrity. SSDs, by comparison, are already operating on microscopic levels, fitting staggering quantities of transistors in nanometer spaces. This is clearly evident by the ever-decreasing size of the modules. Thus, if available space is limited, SSDs can save the system integrator a lot of headaches.

Conclusion
Having read this far, one might think one's choice of storage media is an either-or choice, but this is not necessarily the case. In fact, a combination of the two storage systems may offer the best solution, where HDDs handle data archiving while SSDs take care of any operations requiring high-speed or high-quality recordings for data-analytics. A solid state device might also be worth considering as an addition to already existing setups as it can drastically increase system performance.

While a price gap still exists between the two solutions, SSDs are already more than qualified to enter the market. SSD technology has already shed many of its earlier issues, it can now fulfil the stricter requirements of modern surveillance, and it will expand on what is possible with a well implemented surveillance system.

A graduate in the sciences from Taiwan, Mr. C. C. Wu continued his 14-year career with Flash technology and SSD firmware algorithms for embedded applications. Mr. Wu is responsible for the creation and development of Innodisk's innovative firmware division, and he continues to expand their footprint for Flash solutions. Wearing multiple hats, Mr. Wu worked extensively with the R&D department and sits as the VP of Flash BU and EVP of Innodisk's Branch Office in the US.