Virtualization Technology News and Information
Article
RSS
VMblog Expert Interview: Sunlight Talks Hyperconverged Infrastructure, Project Rosie and Solving AI at the Edge

interview-sunlight-chesterfield 

Last week, Sunlight.io launched its beta program, 'Project Rosie' and with it, the first hyperconverged stack supporting the NVIDIA Jetson edge AI platform, and the Lenovo ThinkEdge SE70, making edge AI deployable at scale.

To find out more, VMblog reached out to Julian Chesterfield, Founder & CEO at Sunlight.io.

VMblog:  What are some of the issues deploying Video Analytics and AI at the edge?

Julian Chesterfield:  Video analytics (VA), computer vision (CV) and AI bring real-time "insight to action" across a wide range of use cases. For example, combining cameras, video streaming and analytics at drive-thrus nationwide for faster and more personalized food ordering; on manufacturing production lines to instantly identify and remove faulty items; and across smart cities to enhance population and crowd security. These sorts of AI applications need high levels of processing power with very low latency and reliable networking in order to give real-time results.

Enterprises want to replicate the simplicity of hyperconverged infrastructure (HCI) they enjoy in their core data centers for their Edge AI applications. However, datacenter HCI isn't able to run in the constrained environments that exist at the edge due to their large RAM and CPU overhead and lack of edge management capabilities. When you have to deploy across 100s or 1000s of ‘edge' sites, the lack of a centrally manageable infrastructure platform makes it extremely resource intensive to manage and hard to scale. This is preventing the deployment of innovative applications, such as AI, at the edge.

VMblog:  How is Sunlight and the NVIDIA Jetson platform uniquely solving some of these issues?

Chesterfield:  NVIDIA Jetson is a fantastic platform for running AI use-cases at the edge. Jetson modules are small form-factor, high-performance computers containing an Arm processor and GPU. They are being rapidly adopted by hardware OEMs for small edge servers that can sit next to a production line or in a fast-food restaurant kitchen.

Sunlight NexVisor is the only hyperconverged stack that is able to run on both x86 and Arm architectures, with a tiny footprint suitable for constrained edge environments. It includes centralized management and application deployment capabilities.

Sunlight was born out of a collaboration with Arm back in 2013 to build a lightweight hypervisor, and we're seeing huge demand for the use of Arm-based servers at the edge for video analytics and AI workloads due to their performance and power-efficiency.

The combination of Sunlight NexVisor and NVIDIA Jetson-based edge servers like the Lenovo ThinkEdge SE70 and Hivecell, makes it possible to run demanding edge AI applications in harsh environments that span 100s or 1000s of sites with easy single-pane-of-glass management, low TCO and tiny power and space requirements. This massively reduces both the hardware and management costs of an edge deployment - making it economically viable to roll out these innovative use cases.

VMblog:  Sunlight recently launched the Sunlight Jetson HCI beta program, nicknamed Project Rosie. Can you tell me more about the program?

Chesterfield:  Sunlight will be giving application developers first access to Sunlight for testing Edge AI applications and a few lucky projects will even get free hardware and the opportunity to be part of the official launch. We will be limiting the program to just 50 participants, which means a waiting list will be in operation. We will likely broaden availability closer to general release (GA).

The types of use cases that are ideal for testing include those with significant security and application isolation requirements (e.g. where the hardware will be in a public place), those that are highly distributed (e.g. across lots of stores or manufacturing lines) and those which require a mixed Arm/x86 infrastructure (e.g. with Arm-based Jetsons pre-processing directly connected video data which is then sent to a larger x86 cluster for deeper processing).

Each applicant will receive a call from a Sunlight consultant, who will assess whether their application is right for the program. Applicants will be able to climb up the waiting list by referring other application developers to the program.

VMblog:  For those considering joining the program, what benefits can they expect to receive?

Chesterfield:  Applicants accepted onto the Project Rosie beta program will benefit not only from the ability to influence the roadmap of the on-going project but also:

  • Access to free Sunlight software
  • Free hardware for some of the most commercially viable projects
  • Free developer support
  • Freebies, merchandise and events (Sunlight's hoodies are legendary)
  • Invitation to participate in the launch publicity later in the year
VMblog:  And can you tell me why you named the program Project Rosie?

Chesterfield:  The Jetsons is an American animated sitcom that originally aired in 1962 and is set 100 years in the future (2062). It has elaborate robotic contraptions, aliens, holograms, and whimsical inventions. The Jetsons are a family residing in "Orbit City" and their housekeeping is done by a robot maid named Rosie.

For this reason, the team agreed the name Project Rosie was most fitting for the beta program - as we're pretty sure she would be running on a hyperconverged edge stack like Sunlight, maybe even on a retro-futuristic version of the Lenovo ThinkEdge SE70.

VMblog:  Finally, how can application developers sign up to be a part of Project Rosie?

Chesterfield:  Application developers can apply to join the waiting list here. We're very excited to see the use cases that ISVs and developers bring to the table, and to help bring them to life at edge-scale. If developers have something that doesn't seem to fit - then we'd still love to hear about it - email us at info@sunlight.io

##

Published Wednesday, July 06, 2022 7:27 AM by David Marshall
Filed under: ,
Comments
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
Calendar
<July 2022>
SuMoTuWeThFrSa
262728293012
3456789
10111213141516
17181920212223
24252627282930
31123456