You’ve likely seen some headlines by now about our recently announced VeloRAIN (VeloCloud Robust AI Networking) architecture. I’ll come back to this in a moment, but let’s jump straight into why it’s important.
During the recent VMware Explore where VeloRAIN was announced, Sanjay Uppal, VP and GM of the VeloCloud division of Broadcom, demonstrated the importance of networking for AI with one example — the impact to the network of Ray-Ban Meta glasses worn by a shopper walking into the store. While the glasses are completely changing the whole shopping experience, the question remains: Is the store network, in this case the WAN, ready to handle this type of new device and traffic patterns?
To answer that, we first need to find out how these glasses work. Sanjay mentioned these Meta glasses are the fastest selling product at the Ray-Ban store today, and I’m lucky to have gotten a pair.
The glasses are super easy to set up. You download the Meta View app, enable Bluetooth, and then the glasses immediately discover and pair with a phone. From there you can start linking different services like Facebook, WhatsApp, Instagram, etc. The glasses primarily use Bluetooth to talk to the phone, but when they need to upload lots of images and videos, they can switch to using built-in Wi-Fi as an uplink for the Meta View app. Otherwise, it is the phone that communicates wirelessly through Wi-Fi or cellular connectivity.
One of the coolest things about these glasses is the multimodal AI interaction with the Meta AI service. The glasses can capture a video or image and then send it to Meta AI through your phone. You can also ask Meta AI to describe these images. However, current capabilities are somewhat limited; for example, it cannot yet tell me the price of an item that I am looking at.
Meta seems to indicate that more is coming. For instance, while testing these glasses at an airport duty-free store, the glasses immediately responded to my request by taking a picture and, within seconds, playing an audio response from the Meta AI service. Below is a screenshot of the Meta View app showing my interaction with the glasses.
In addition to this multimodal AI use case, it can also do video calls, capture video and post to your Facebook or Instagram, or send a 15-second video clip through chat services such as WhatsApp.
These Ray-Ban Meta glasses are very powerful. They have the latest smart glasses processor with integrated AI capabilities for inferencing, image and audio enhancement, an integrated camera capable of capturing 12MP or QHD (2K) video — which provides better resolution than cameras in most modern laptops — and 32GB of storage. I am amazed there is so much tech packed into something that weighs less than a piece of toast.
Back at home, I became curious about how the glasses actually work, so I connected my phone with the Meta glasses behind a VeloCloud SD-WAN Edge and started running some traffic tests. The results were eye-opening:
In short, the glasses provide a simpler way for users to interact with Meta AI service, similar to what you can do through WhatsApp today. However, the glasses make it much easier. All I need to do is to say “Hey Meta, what am I looking at?” The glasses immediately take a picture, upload it to Meta AI, and deliver an audio response almost immediately. So it is not surprising that when we take a look at the traffic generated by the Meta glasses, you see a significant upload to download traffic ratio — almost 8:1. The chart shows volumes in KB, so this one pair of glasses generates or requires approximately an average of 4 Mbps of WAN bandwidth to upload just the picture.
Why should an enterprise — or in this case, why should a retailer — care? Most retail stores today provide Wi-Fi service, out of necessity because shoppers expect it. Some shoppers rely on Wi-Fi because of a lack of good cellular connectivity or dead spots in the store. Retailers want to bring shoppers back into their physical stores, so ensuring a great shopping experience is paramount.
These Ray-Ban Meta glasses will for sure put more load on the enterprise WAN, but they also promise to deliver a revolutionary shopping experience.
- Shoppers may want to look up reviews or compare prices of similar items at the stores. These glasses will drive multimodal interaction with the Meta Al (and perhaps others as Meta expands its capabilities). Typically, guest Wi-Fi is provided with best-effort service, but retailers may want to rethink that. Identifying this traffic and being able to provide the right treatment, like proper prioritization, is important.
- Influencers or shoppers may want to stream what they see live from the stores to their followers and their friends. Again, the ability to identify and accommodate this traffic dynamically or on-demand during certain times of day or critical events will be important.
On the flip side, enterprises concerned about security may decide to not allow the traffic from these glasses at all. Imagine your employees with these glasses walking around either knowingly or unknowingly taking pictures of whiteboards or capturing sensitive information. Having confidential information showing up on the Internet is a scenario enterprises would want to avoid. Organizations will need robust methods to identify and manage such traffic in line with their security policies, whether that involves blocking, auditing, or monitoring it.
This is where VeloRAIN comes in.
We developed the VeloRAIN architecture to enhance how enterprises build and manage AI-driven networks, with smarter application and traffic prioritization — to help manage these shifts in traffic patterns and new demands of AI workloads. VeloRAIN not only guides the inclusion of AI in VeloCloud products, it provides the networking structure needed to handle the rapid growing volume of AI applications, especially with the proliferation of AI at the edge.
At VeloCloud, we are working to ensure our AI-powered architecture, enabled by VeloRAIN, delivers the visibility, control, and security needed for managing the next generation of AI-driven traffic—like that generated by Ray-Ban Meta glasses. For more information, see our blog VeloRAIN: AI Networking is the Next SD-WAN Revolution.