Page 42 - Security Today, September 2019

best locations to place food carts. This doesn’t even cover using ana- lytics for security reasons. In a specific frame of video, is that person holding a gun? What is the license plate on that car that sped away?
Buildings, facilities and cities are becoming smarter and more connected. Although the biggest server workload within most se- curity departments is driven by video recording, now applications as diverse as access control, lighting and HVAC have all become digital. Access control, lighting and other systems are based on In- ternet-connected devices–sometimes referred to as the “Internet of Things (IoT)”–and require oversight and management from the se- curity and facilities department. Some examples include being able to adjust lighting for each office in a way to create a better working environment, or adjusting the angle of window shades throughout the day to maximize sunlight.
These major trends have combined to drive the need for hypercon- vergence in surveillance and building systems.
What is Hyperconvergence?
Hyperconvergence is a term that originated with large cloud pro- viders and large enterprise IT departments. It refers to “converging together” separate compute/servers, networking and storage into a single hyperconverged infrastructure.
To understand hyperconvergence, let’s start with what most se- curity department infrastructures look like today. A typical security infrastructure utilizes a traditional “three-tier” architecture, which simply means they have:
• Servers that run applications such as VMS software or an access
control application, etc.
• Networking switches that connect the servers to the storage.
• Storage systems where surveillance footage is retained.
New software technologies have emerged in recent years that en- able companies to hyperconverge these functions onto a single server platform. This server platform contains embedded storage, and smart software handles the communication and networking functions.
An architecture based on this hyperconverged design is referred to in enterprise IT departments as a hyperconverged infrastructure, or HCI. This architecture has been widely deployed for enterprise applications in the datacenter but it has not yet been widely adopted for surveillance and security.
Figure 1 – What is Hyperconvergence?
In comparing the traditional infrastructure with the hypercon- verged infrastructure, it is helpful to break down how the data trans- fer occurs in either scenario by looking at a VMS recording server.
In the traditional model, the VMS recording server is connected to a camera over the network. The camera sends data over the network to the recording server, and the recording server then sends that data to the network switch, which then sends the data over the network (again) to the storage device. Each step adds some network latency.
In the hyperconverged model, the VMS management server, as well as the recording servers, runs directly on the same server that is controlling the storage. Surveillance footage is sent over the network,
but the VMS application then communicates directly with local stor- age, bypassing the need for an intermediate network switch. This ap- proach reduces latency and is more efficient.
Hyperconvergence: Why It Matters for Security and Surveillance
I know what you’re thinking; hyperconvergence is a “buzzy” technol- ogy term–used widely in the cloud and in enterprise IT. Why does it matter for surveillance and security?
First, it dramatically simplifies the architecture and makes it easier to manage. Rather than manage multiple servers, switches and storage devices, security and facilities professionals can interact with a single platform and with fewer components requiring maintenance and management. This makes their lives easier, but also improves security and reduces business risk by reducing technical complexity.
Second, one of the attributes of a hyperconverged infrastructure is that the “compute,” or server resources, sit right next to the storage. This becomes critically important when running analytics against frames of video. Analytics will run faster on a hyperconverged plat- form, and this can lead to discovering potential security threats and vulnerabilities more quickly. It could also translate into discovering new ways to optimize a business, such as managing staffing and more.
Hyperconverged architecture is built for the smart buildings and smart cities of today and tomorrow. Rather than managing these applications on different dedicated servers–which is expensive, inef- ficient and complex–these applications can now be hosted and run on the same hyperconverged server that is capturing video.
An Inflection Point for AI and Machine Learning
We are at an inflection point in this market where surveillance is tran- sitioning from its historical role–which was strictly security and loss prevention–to also providing data that can be used to gain new in- sights and business opportunities. At their core, many of these appli- cations are using compute-intensive analytics to, for example, analyze a large number of frames of video to search for objects and trends. These compute-intensive operations can be run on a central process- ing unit (CPU) and now a graphics processing unit (GPU).
A major advantage to hyperconvergence is that it “enables” these types of analytics because the compute resources are close to the storage. These architectures can be more flexible about deploying
CPU and/or GPU resources to suit the needs of the application. However, one of the downsides of a hyperconverged infrastruc- ture is that the traditional model allows compute and storage re- sources to be scaled independently. This can prove important if, say, a security department wants to increase surveillance retention times. That change may not require any additional compute resources, but
it does require additional storage.
Conversely, if a security department is adding more cameras and
already has sufficient storage in place, new compute resources can be added independent of storage. A hyperconverged infrastructure has fewer “building blocks” and is simpler in that respect, but one implication is that compute and storage resources are always added as the system scales. Many HCI vendors minimize this disadvantage by offering different hardware configurations that can be adjusted to meet the specific needs of the user.
Considerations For HCI Solutions
There are many server and storage vendors that sell hyperconverged solutions into the surveillance and security space. Many of these hy- perconverged infrastructure products were designed to run enterprise IT workloads and applications but were not necessarily designed for the unique workloads video surveillance can produce.
The main workload within a security infrastructure is still video
40
0919 | SECURITY TODAY
SECURE STORAGE