The dramatic explosion of video data expected over the next several years is a staggering testament to the hyper-convergence of myriad IoT technologies. It is predicted that video will account for 15.1 zettabytes of data annually, which is more than any other IoT application. This influx in data, as well as the growing adoption of video analytics and AI, propels video to be used in a wide variety of ways. These trends enhance video’s value but also pose challenges, such as threats related to cybersecurity.
At the recent SIA Cyber Secured Forum in Dallas, Texas., Brandon Reich, Senior Director of Surveillance Solutions for Pivot3 provided some insights into this new world of IoT and its expanding role in video surveillance and the cybersecurity challenges the industry will face as data is stored and shared across the enterprise. Pivot3 will be showcasing several of its video surveillance solutions this week at GSX in Chicago on September 10-12 at booth #2654, including the Pivot3 Acuity IoT-Surveillance Series, which is purpose-built for demanding, data-intensive video workloads of any size.
SIW: As the IoT universe expands, where does video surveillance fit into the picture?
Brandon Reich: What we're concerned about is just the explosive amount of growth that we've seen in data generation, particularly from video surveillance and the implications that it has around how we secure, protect and ensure the ongoing viability of those systems. Those are really the two key things that we're most concerned about and what I want to share. But we're going to do it in the context of this broader trend that is called IoT or Internet of Things. For the most part, video was really the first original connected device; you can say it was the beginning of the IoT revolution going back decades.
Obviously, video surveillance is not a new technology. It's been around for decades. But the demand for video surveillance is at an all-time high. There are a couple of reasons for that. One is we know there have been significant advancements in video technology. Video quality and bandwidth issues have improved drastically. We've seen that there are more use cases for the technology now. We're not just using video for the traditional case of catching the bad guy. We can use it for business and operational issues, as well as other non-security functions video surveillance. Of course, the cost of video technology has come way down as well.
All of that has led to a huge demand for video and that has resulted in video increasingly coming under the auspices of, or under the control of, IT in our organizations. Oftentimes, at least the infrastructure part of it is moving into the data center, and with that comes different kinds of technologies and different kinds of expectations around how we use, manage and secure all those technologies. We have to keep that in mind. But video is certainly not alone in the world of IoT. There are dozens, if not hundreds and thousands of different types of sensors out there being deployed across our facilities, across our cities, really across the world and all used for different purposes. Oftentimes, they're used together. Sometimes they're used separately.
But all these various sensors share one thing in common and that is that they're l generating data of some sort. That data is being turned into intelligence. We can turn information that a sensor is giving us into something smart for us, right? But what we don't always think about are the security, the information, the data security implications of all these different devices, all these different sensors that are being deployed in so many different aspects of our world.
SIW: Can you provide a picture of what the role of video devices are currently playing in this world of IoT?
Reich: We're starting to see a little bit of the IoT industry characteristics overlap into the physical security world. But I want to take you through some more of these numbers in a little bit more detail and granularity. I think you might find this interesting. Gartner estimates that this year 2019, one billion new IoT devices will be shipped and installed. Of that one billion devices, about 10% of them or so will be video surveillance cameras. So, roughly about a hundred million video surveillance cameras will be shipped and deployed this year. However, while those cameras only represent 10% of all devices, those 100 million security cameras are responsible for generating 84% of all the IoT data in the world. Think about that -- 10% of the devices are responsible for 84% of the data.
If you step through that even just a little bit further, you'll see that video surveillance is responsible for generating about a third, 36%, of all the data in the world, not just IoT data but all data. Video surveillance is generating massive amounts of data. IHS now estimates that for the first time in our history, over 100 million new cameras will be shipped this year, which aligns with Gartner’s projections.
The fact that the average resolution of the camera being deployed today has gone from 720p, just about two and a half to three years ago, to close to three megapixels now creates another challenge. These cameras are generating much more data as a result of their higher resolution. Wikibon, which is IT analyst firm, said that in 2019 video surveillance will generate almost 39 zettabytes of new data. That's up from 15 zettabytes two years ago. Thirty-nine zettabytes of new data. That means that video surveillance is the largest generator of new data in the world by a factor of almost 2X. What's more, is that the rate of new data generated by these cameras is increasing at about 42% per year.
You see the implications now. We're generating massive amounts of data. That data, because we're using it for so many different purposes and with so many different technologies, that data is much more important to us than ever before. Conversely, the need for us to protect and secure that data is at an all-time high as well.
SIW: What makes video data so different from other forms of data that is stored and disseminated through an enterprise data center?
Reich: There are a couple of things that we need to think about when it comes to deploying infrastructure technology for video. The first one is performance. When I say performance, I mean the infrastructure system's ability to take in or to ingest all that video data being generated by all those cameras without any kind of loss. Loss in the video world is a really bad thing. What's interesting about this is a general rule of thumb for most IT applications out there. It is said that most IT applications spend 20% of the time writing data to a disc, while 80% of the time they're reading data from a disc. As a result, most IT infrastructure technology is optimized for that sort of 20 to 80 ratio. But video surveillance is the exact opposite. Video surveillance is more like 95% to 99% of the time writing data to a disc, while it spends a small fraction of the time reading it.
Video is also a highly variable workload. Which means that the amount of data coming out of those cameras can change very quickly, often in a very short period. When you put those two things together and try to ingest data from dozens or hundreds or thousands of cameras, you can quickly overwhelm your infrastructure's ability to take all that in. That leads to data loss. Data loss in the video world is something called frame loss, and frame loss is a really bad thing. It can lead to either degradation of the quality of the image or it can lead to the loss of entire segments of video depending on which frame you lose. We must think about infrastructure that can perform in a way to capture all that video without any kind of loss.
The second thing to think about in infrastructure deployment is resiliency. When I say resiliency, what I mean is protecting your systems against downtime and data loss that might occur as a result of major hardware failures. Hardware will fail. In fact, hardware used in video surveillance is about three times more likely to fail than that same hardware used in non-video applications. That really goes back to that first issue related to constantly writing data to disc. In fact, Google put out a study a couple of years ago that proved a hard drive is three times more likely to fail when it's used for video than when it's used for anything else.
SIW: So, what is needed to guard against this issue?
Reich: We must look at how do we protect our systems against downtime and data loss when those failures occur. Most of our IT counterparts would say, you know, "We have lots of great technology for resilience. We have things like replication and snapshots and clones and cloud and backups and all this stuff." My response to that is typical, "Yes, that's fine. However, when you use those things for video they can either be prohibitively expensive," because it's very, very expensive to replicate hundreds of terabytes or petabytes worth of data, or, in many cases, a lot of these technologies just don't work. You can't use snapshots in video systems.
In some cases, the cloud becomes prohibitively expensive. We know about the challenges with cloud in large scale mission-critical systems. Backing up data technically can be done, but if you've ever done that and tried to recover the data, it can be a real challenge. We have to think about other methodologies, other ways to protect our systems against downtime and data loss because hardware failures will occur. And while it's not necessarily related to protecting data or protecting video, we also must consider scalability when dealing with infrastructure deployment.
I think the worst kept secret in the video industry is that video systems over time never get smaller. In most cases, they almost always get larger and there are a lot of different reasons why that happens. However, there's a common misperception out there that as your video system gets bigger, you simply scale your storage capacity to keep up. While it's true, you do have to scale storage, you also must scale compute and bandwidth capacity and you typically have to do so linearly, meaning together all at the same time. You always want to do it in a manner that's non-disruptive to your existing applications. Performance, resiliency, and scalability are very critical things when it comes to planning infrastructure deployments for large scale mission-critical video surveillance.
SIW: What are you seeing as some of the latest technology trends across both the IT and physical security bridge?
Reich: About a decade ago, many organizations started looking around all of their functions and said, "How can we reduce costs? How can we reduce costs and complexity in our organization?" IT was no different. IT started looking very closely at their data centers. They discovered data centers were massive cost centers. They were built on big expensive, complex hardware, proprietary technology. It cost them an absolute fortune to set up, deploy and to administer. Some of the leading IT innovators in the world looked for a different way to do it. What they came up with was what's now considered the concept of a software-defined data center.
Now, what that really means is companies like Amazon, Facebook, Google said, "Let's get rid of all this big expensive proprietary hardware that we have in our data center. Let's just use this low cost, off-the-shelf commodity hardware, and let's replicate all the functionality of the expensive stuff with software. That's what they did; they went out and built their own data centers, built their own software and were able to vastly reduce the cost and complexity of administering their data centers. Well, that's great for companies like Amazon, Facebook, and Google who probably have a couple thousand IT people apiece in their organizations. But what about everyone else? That led to the emergence of an industry or a data center technology category called hyper-converged infrastructure, which is really the same concept of software-defined data center solutions purpose-built for a specific type of application. This is one of the big trends happening in the IT world that has become very relevant here to us in the physical security world.
SIW: Why is the hyper-converged infrastructure approach gaining steam?
Reich: Hyper-converged infrastructure has evolved to the point that it's now very commonplace. It's very cost-effective and it's very usable in and around physical security technology industry. Hyper-converged infrastructure starts with low cost, off-the-shelf commodity server hardware that in many cases uses the exact same hardware employed by NVRs. Then it leverages software, some that are specialized, to turn groups of those servers into enterprise-class data center infrastructure that can be optimized specifically for a workload or an application. In this case, for video surveillance.
Some of the high-level benefits you can achieve include much higher levels of resiliency, eliminating downtime, eliminating data loss that might occur when hardware failures happen, which are inevitable. You can achieve and sustain the highest levels of video performance and eliminate frame loss. Since it is all software-defined, you can tune this infrastructure specifically for a video surveillance application.
You can also scale it by starting small and then scaling bandwidth, storage and compute together over time as your needs and your budgets change. Most importantly, you are able to reduce the cost and complexity associated with this kind of enterprise-class infrastructure for physical security now, and you do that by having the ability to consolidate multiple different workloads, multiple different technologies, video surveillance, access control, analytics, visitor management -- whatever your applications are. It can eliminate the need to buy servers and hardware for all those different systems and consolidate them while being able to administer all of it through a nice single pane of glass. This concept of hyper-converged infrastructure brings the benefits of enterprise-class infrastructure technology without the cost and complexity typically associated with that.
About the Author:
Steve Lasky is the Editorial Director of SecurityInfoWatch Security Media, which includes print publications Security Technology Executive, Security Dealer & Integrator, Locksmith Ledger Int’l, and the world’s most trafficked security web portal SecurityInfoWatch.com. He is a 30-year veteran of the security industry and a 27-year member of ASIS. You can contact him at firstname.lastname@example.org.