Overcoming the storage challenges of adopting surveillance AI

Overcoming the storage challenges of adopting surveillance AI
From small businesses to large enterprises to smart cities, entities are relying on data to improve operations, safety, and the user experience. Thanks to AI, integrators and solution architects are using IoT devices, surveillance cameras, and other sensors to capture data in more creative ways. However, this data is rendered useless unless it is securely recorded, analyzed, and delivered for actionable results.

Data is in flight all around us

Data is in flight all around us and has become an essential part of the human experience. IDC forecasts the global datasphere will increase from 33 zettabytes in 2018—where one zettabyte equals a trillion gigabytes—to 175 zettabytes in 2025. The report says, “Every connected person in the world on average will have a digital data engagement over 4900 times per day.”

Executives are ultimately looking to interpret the data aggregated by IoT devices, sensors, and security solutions and leverage it to improve operations, cost-savings, and customer satisfaction. The deployment of cognitive systems—such as machine learning, natural language processing, and AI—that actively analyze this data for proactive decision-making are on the rise. IDC indicates that “the amount of analyzed data that is ‘touched’ by cognitive systems will grow by a factor of 100 to 1.4 zettabytes in 2025.”

 Storage in the era of AI

New enhancements allowing security solutions to be used for business intelligence is driving the demand for data-hungry applications. The increased use of AI systems in security has warranted a shift in recording and storage technologies. Standard surveillance systems primarily recording footage were typically write-only applications. Today surveillance systems with AI have mixed read/write workloads.

To remedy this issue, storage providers are building AI into video NVR systems and harnessing the power of micro-datacenters so that initial processing, analysis and pattern recognition occurs in real time at the edge

Development of AI-enabled NVRs and edge computing devices are driven by cheaper graphics processing units (GPU) with enhanced analysis capabilities, as well as better storage options. In particular, new hard disk drives with fast writing data speeds, high read performance, and support for both AI and video workloads have become attractive solutions for system integrators.

After the initial video ingestion and analytics at the edge, video is pushed to the back end or cloud. In this centralized environment, video and AI metadata are consolidated for deep learning activities to train the system to be more predictive and provide a more holistic view of the video data collected

Smart city sectors

The development of safe and smart cities continues to be one of the sectors where surveillance systems and data will have the greatest impact. Research firm IHS Markit indicates the global market for city surveillance exceeded US$3 billion in 2017 and is expected to increase each year by 14.6 percent from 2016 through 2021.

Beyond citywide surveillance, smart cameras, IoT sensors, and edge computing devices with AI are being deployed in smart cities to equip businesses and citizens with data that can enhance the urban experience. In hospitals, IoT sensors and video devices enable remote patient monitoring, providing real-time alerts.  Intelligent traffic lights allow for optimized routes based upon mobility patterns. Smart energy usage meters allow for more frequent readings and energy consumption tracking, providing citizens with information that can be used to facilitate lifestyle adjustments that generate cost savings.

Storage considerations for data hungry applications

To accommodate this deluge of data collection for smart cities and other AI deployments, system integrators and solution architects need to ensure they have the proper storage configuration. Otherwise, customers will be subject to dropped frames and data loss, which inhibits deep learning and predictive analysis. Here are the top storage best practices for security professionals.
  1. Implement IT 4.0 Architecture – In this era of unprecedented data collection and analysis, a new storage architecture is needed. Because of the plethora of sensors used for advanced applications, integrators can no longer rely on the cloud alone as a viable or even cost-effective solution. Instead, integrators need to deploy high-performing storage solutions at each stage of the data flow, from the endpoints, edge, and cloud. This is referred to as IT 4.0. For more information, refer to the IT 4.0 Storage Framework infographic.
  2. Select the Right Hard Drive – Not all hard drives are created equal. When it comes to powering your DVR, NVR, or server, choose a hard drive that is purpose built for surveillance. Standard desktop drives are best suited for applications that operate 8 hours a day, 5 days a week. This is not the case for surveillance systems, especially for heavy analysis use cases. For AI and smart city deployments, cameras are recording 24×7, and storage systems often have longer retention periods for deep learning. Integrators would do well to utilize a hard drive like Seagate’s SkyHawk and SkyHawk AI, which are seventh-generation products optimized for surveillance with 3× the workload of typical desktop drives. They record up to 64 HD cameras—storing 10,000 hours of video—and feature ImagePerfect firmware to improve streaming. SkyHawk AI also supports 32 streams of AI metadata. Built-in rotation vibration sensors ensure SkyHawk and SkyHawk AI perform optimally in 16+ NVR bay environments. For centralized storage locations that need to scale to petabytes of data from thousands of cameras, Seagate Exos enterprise-class drives features higher capacities, SED and cybersecurity features. Seagate Nytro solid-state drives also improve speed and performance for blade servers managing hot data.
  3. Employ Drive Monitoring Software – To ensure the longevity of the overall surveillance and storage system, select hard drives with embedded monitoring software, such as SkyHawk Health Management (SHM). Built into SkyHawk drives, SHM provides continual updates on how the drive
  4. Enroll in Data Protection Services – For additional peace of mind, integrators should opt into data recovery services. In the event of a power outage, vandalism, equipment malfunction, or natural disaster, customers can still retrieve their data for up to two years with Seagate’s Rescue Data Recovery Services plan.

The collated data is rendered useless unless it is securely recorded, analyzed, and delivered for actionable results. It is critical that robust storage systems are implemented from edge to cloud for intensive recording and heavy analysis applications.
By implementing the best storage architecture and technologies, businesses can leverage data in real time to respond more quickly to high-risk scenarios and make smarter decisions that have a tangible impact.


Share to:
Comments ( 0 )
security 50