Technical Walkthrough

Accelerated Edge AI with Metropolis and Fleet Command

Discuss (0)

Approximately 1 billion video cameras—the ultimate Internet of Things (IoT) sensors—have been deployed throughout the world’s cities and spaces to help us live better and safer. Optimizing AI-enabled video analytics is critical for frictionless retail, streamlined inventory management, traffic engineering in smart cities, optical inspection on factory floors, patient care in healthcare facilities, and more. 

NVIDIA Metropolis is an application framework, set of developer tools, and partner ecosystem that brings visual data and AI together to improve operational efficiency and safety across a broad range of industries.

But the journey to AI at scale is hard.

From defining requirements, procuring hardware, developing and fine-tuning software, IT and security reviews, to physically installing and updating system and application software at each site, the process is arduous, time-consuming, and costly.

Developers operate best when they focus on building AI applications, and are not burdened with the responsibility of supporting the hardware, security, and scalability issues that can arise when deploying AI at edge locations.

Deploying starts with the pitch, and moves to determining requirements and acquiring hardware, installing and tuning software, and security and IT reviews. The end stage is broad installation and ongoing updates. This process can take months to complete.
Figure 1. The typical timeline for AI deployment can take months from POC to scale.

NVIDIA Fleet Command addresses these issues and makes it easy to deploy AI applications. Fleet Command is a turnkey cloud service that allows administrators of any skill level to manage the AI lifecycle by securely deploying, managing, and scaling AI across distributed edge infrastructure all from one control plane. Remote orchestration of AI applications across devices at the edge means that Fleet Command can save organizations weeks of planning and manually implementing edge infrastructure. 

Fleet Command questions

Recently, we hosted a Metropolis Developer Meetup to show how Fleet Command can be used to deploy vision AI applications at the edge. The webinar can be viewed on-demand.

With hundreds of developers looking to move their applications from POC to scale, there was great discussion and engagement from attendees. Below are some of the top asked questions with our answers.

How does Fleet Command help speed up the deployment of AI applications?

Fleet Command is a turnkey solution that removes the complexity of building and maintaining an edge software platform:

  • Simplified Edge AI: Fleet Command removes the complexity of building, operating, and maintaining an edge software platform, offering administrators an “IT free” tested and optimized solution that allows them to move from software installation to edge deployment in just a few clicks.
  • Streamlined deployments: Administrators of any skill level can deploy and scale applications to any number of locations. One-touch provisioning turns systems into AI appliances and eliminates the need for skilled IT technicians to travel to edge sites to install servers, accelerating the time to AI insights.
  • AI lifecycle management: Update applications over the air, easily scale additional applications, and monitor AI health from system to applications. A resilient software platform automatically restarts applications and migrates workloads at the onset of problems. Detailed monitoring dashboards and extensive automated processes make managing AI easy.
  • Layered security:  From cloud to edge, leading security protocols ensure application data is always protected, shifting the burden off organizations to build and operate these features. A secure private registry isolates intellectual property and encryption in transit and at rest safeguards valuable data.
The image shows the Fleet Command interface, which allows users to manage multiple edge locations with one interface. There are three easy steps to use Fleet command: load your application, set up an edge location, and then connect the edge device to the software. Users can then manage their edge locations with Fleet Command and use NVIDIA NGC to deploy pretrained models.
Figure 2. With Fleet Command, your edge fleet can be centrally managed with one interface.

What requirements do I need to meet to deploy my application on Fleet Command?

Fleet Command deploys containerized applications on Kubernetes through Helm charts. To learn more, see the Fleet Command application developer guide.

What metrics can Fleet Command track? Are there system metrics or can data be pulled from the application?

In addition to system and application logs, Fleet Command also provides alerts for system and location health and deployment status.

Do I need to use the Fleet Command user interface or can this be done using APIs?

Fleet Command APIs can be leveraged by partners to integrate into their existing solutions to provide organizations deploying edge AI, a single consolidated edge computing platform.

What happens to servers when they lose Internet connection? Can Fleet Command still operate?

Yes, applications deployed on Fleet Command can still operate without an Internet connection. When disconnected from the Internet, servers at the edge continue to operate and provide insights. The only functionality that is lost is the ability to make updates or control of those systems until the Internet connection has been reestablished between Fleet Command and the edge systems.

What hardware does Fleet Command support?

Fleet Command requires the use of NVIDIA-Certified Systems at edge sites. This is a comprehensive list of all NVIDIA-Certified Systems. However, it is recommended to deploy a system that is designed for edge use cases.

For general guidelines on configuring the specifications of your system for your requirements, see the NVIDIA-Certified Systems Configuration Guide.

How can I see a demo of Fleet Command? 

Watch this demo to see how easy it is to deploy AI across their edge infrastructure in minutes with Fleet Command.

Learn more about the power of Fleet Command and how it can help you deploy, manage, and scale AI at the edge by watching this on-demand Metropolis developer meetup session, featuring use cases presented by Data Monsters.

Also, join us at NVIDIA GTC to discover more about the power of edge computing and Fleet Command. The virtual event is taking place November 8-11. Make sure that you register today.