High-Volume Storage

June 30, 2009

The demands on video surveillance systems have grown dramatically in recent years. Improved video quality, higher frame rates and the need for longer retention periods for recorded video have led to a significant change in the storage that video surveillance deployments require.

While most security practitioners focus on the capabilities of the surveillance camera or the VMS, the biggest item in any modern deployment is often storage. In fact, 1/3 to 1/2 of the rising cost of surveillance typically is from this decidedly unglamorous component.

The laws of probability related to disk drive life spans and actual history both reflect that about 50 percent of DVRs can be expected to experience failures and video loss during their lifetime. To address this reality while reducing costs, some storage vendors now offer external DVR upgrades. The best of these support even small CCTV systems with just one or two DVRs for less than the cost of adding an additional DVR. They also eliminate the risk of lost video, support the maximum frame rate/resolution provided by the attached cameras and DVRs, and extend the recording capacity of multiple DVRs.

Growing Storage Needs

Nowhere is there more demand for high-volume storage than in the growing IP world. Megapixel cameras, video analytics and video management systems have resulted in dramatically improved video imagery with a requirement for longer retention periods at a higher resolution and frame rate.

Most security practitioners have addressed this by one of two approaches: They purchase commodity off-the-shelf application servers to host a VMS, and plug storage into that commodity server; or, they opt to go with the newer concept of shared, external IP storage. Let’s look at how these approaches and the storage used differ.

First, not all storage is created equal. Just as cameras differ in performance and capabilities, so does storage. Those coming from the security world often think of storage simply as commodity disk drives inside or external to DVRs, NVRs and servers. Those with an IT background typically think more in terms of shared, clustered storage savings and benefits common to data centers. Both views have merit, but neither is complete.

Most DVRs and many NVRs feature consumer- or PC-grade disk drives. They are designed for typical desktop applications like word processing, e-mail and Web browsing. More expensive are server- or OEM-grade drives, designed for more demanding workloads and with increased reliability and longer expected service lives without failures. These are the drives typically found in IT storages in data centers, but not in DVRs.

But the real difference in storage is not the disks, but in overall design. What the storage system is designed to do, how it is managed and how it is used are each much more important than the individual components.

The largest user of storage systems today is IT. Data centers employ skilled IT storage professionals to design, administer and maintain their storage systems. Frequent, scheduled system downtime is required to allow maintenance such as applying software updates, tuning performance and defragmenting the storage itself — and many are designed to operate only in climate-controlled data centers.

Unfortunately, this is not often the model in security. Security practitioners do not want to become storage experts, nor do they wish to frequently bring their systems down for maintenance. Instead, the best video-optimized solutions are designed to setup quickly and easily in just a few minutes, and then to run unattended with e-mail notification of system problems. And instead of a large data center with environmental controls, security systems are often found in cramped security rooms, tight wiring closets or on/under desks.

By far, the biggest difference is in workload. IT systems are typically generalist solutions, designed to run a balance of read (from disk) and write (to disk) activities for applications such as databases, human resource management, sales, payroll, engineering, manufacturing, etc. They typically run in batch (all at once) or online transaction processing (such as database updates), and multiple applications typically run on the same computers and storage.

Video surveillance workloads differ greatly. They typically run a single application flat out, recording (writing to disk) 95-98 percent of their operational time. Only 2-5 percent of the time is spent on playback (reads from disk), and that typically is only when an incident occurs. To perform at their best, video retention systems need to be tuned for near-constant recording. Unfortunately, this unusual and demanding workload challenges IT storage simply due to design differences.

The solution is to select a storage product that delivers the power of shared IP storage from the IT world, but that delivers a design specifically for physical security as the primary application — not an add-on.

Security users often deploy commodity servers to host VMS systems and support direct attached storage. Unfortunately, doing so will usually require more hardware than opting for a shared, external IP storage solution.

Consider the example of a major municipal electrical utility in Texas. The requirement was to support three different brands of IP and megapixel cameras providing coverage across a large geographical area and multiple facilities. Starting with 90 cameras and expected to grow to 300, the utility selected one of the best known VMS vendors. Sizing the hardware, they found they required 18 commodity servers with a total of 128TB raw capacity. The cost was stunning, nearly a quarter of million dollars.

By opting to instead use shared IP storage, enormous savings were achieved. The number of servers shrank by nearly 90 percent — from 18 to just two (see figure 1, page 40). And that meant the storage could be reduced by nearly 80 percent, with just 30 TB required. Reliability was dramatically improved, with OEM-grade drives and components designed into the video-optimized solution, eliminating downtime and the risk of lost video. The total cost reduction of about 70 percent cut the price to just $70,000. The green benefits of reduced power consumption — less to heat and cool, and much less hardware to manage and maintain — added even further savings.

Outside of physical security, large storage requirements are nothing new. In IT, storage capacity has been growing for decades, with succeeding waves of technology to better address requirements. Like today’s DVRs, the first IT solutions used direct attached storage (DAS) captive to a particular application server. This was replaced by the idea of sharing storage to improve performance, benefit from quantity of scale and improve reliability. Network Attached Storage addresses this for IT workgroups and departments, although performance, capacity and administration can be challenging when users try to grow into larger deployments. Instead, Storage Area Networks are commonly used for higher-performing, large-scale needs.

Originally, fibre channel SAN was the only choice, which restricted use to large data centers. When less expensive but equally scalable and reliable IP SAN came along, it was possible for customers that previously could only afford to deploy DAS or NAS to benefit.

Emerging Choices

The example illustrates how using IP SAN benefits security users over DAS and commodity servers. Yet there is still another technology wave coming from IT that is making itself known in video surveillance — virtualization.

Virtualization is a simple concept. Instead of running two or more applications on separate hardware platforms, virtualization enables organizations to host the applications simultaneously on a single, larger server. It cuts costs, improves reliability, reduces maintenance and dramatically “greens” a system by reducing power consumption, heating, cooling and equipment needs.

Going back to the earlier example, 16 of 18 commodity servers were cut with cost savings approaching $200,000 for hardware alone. What would be the impact if we could cut out those two remaining devices?

We can do it by taking the two commodity servers that each run a copy of the VMS, and hosting the applications as virtual servers on the shared IP storage platform. This cuts the cost of hardware further, along with less energy, heating, cooling, etc., all over again (see figure 2 above).

Virtualization is now here for security, from a small number of vendors. The best of these approaches can start with as little as 2TB of capacity, yet can scale modularly as needed so application needs can start affordably and small, yet grow when needed. Not only does this save money, but is very simple to deploy without complexity.

Selecting the Best Solution for Your Needs

Speaking of complexity, it is actually pretty easy to determine if you are being offered a video-optimized solution designed for physical security, or just a re-purposed IT platform that will not meet expectations over the long term. Remember, security is not a science project. We protect people and property, and much more — make sure your storage choice is designed with that in mind.

Here are four simple questions that can lead you down the right path of product selection:

1. Does the system require IT expertise to keep in running? If it is expected that IT personnel will maintain the storage solution, it is not designed to meet the needs of physical security workloads and environments, but is just repurposed hardware. Be very suspicious.

2. Can it start small and grow, or must you invest in a large system immediately to get all the benefits? The best solutions can start with as little as 2TB of capacity, and grow modularly as needed without ever halting recording. Otherwise, you are paying for more than you need.

3. Is it certified with multiple physical security applications, and is the vendor engaged with industry associations like ASIS, NRF,USGBC, SIA and ACA? Video workloads and security requirements differ from those of IT dramatically. Only those who have taken the time to test and certify with security applications know the problems, and how to solve and avoid them. Unless the vendor participates with industry associations that matter to security, how can it really know how to meet your needs?

4. Do you need to be an IT expert to setup and install the system? The best systems set up easily in just minutes, and use an interface that talks in physical security and not complex IT terms. If you need IT help to set up even the smallest systems, it was not designed for security users and you will uncover lots of other problems going forward. Pick something else.

Thinking these issues through will enable security users to take advantage of the best on the market, without falling victim to overly expensive or complicated solutions. You will benefit with more reliable, more expandable and easier-to-maintain systems. And you will save money to boot.

Jeff Whitney is vice president of marketing for IP storage vendor Intransa. He offers extensive physical and network security knowledge, combined with IT storage expertise developed over 20 years as an industry practitioner. He is a volunteer on the ASIS Intl. Physical Security Council, and is the primary Intransa representative to the Security Industry Association.