How Edge and Cloud Computing are Spearheading the Retail Digital Transformations
By Motilal Agrawal - April 01, 2021
I’ve been wanting to do a more detailed blog post comparing edge and cloud computing, and how Zippin technology is taking advantage of advances in both, and edge computing specifically. Before doing a deep-dive, let me take a moment to summarize what the cloud has to offer.
The cloud offers potentially unlimited raw computing power with the added advantages of:
- Efficiency: These data centers can operate at high power efficiency thereby being extremely efficient in power and thermal dissipation.
- Flexibility: When the compute demand is unpredictable or highly variable, the cloud offers the flexibility to elastically scale the compute required to get the job done.
- Economical: The ability to manage the infrastructure remotely saves on the IT maintenance costs for maintaining these infrastructures.
All of these advantages have led firms to abandon their in-house data centers for the cloud. While these are certainly compelling reasons to migrate to the cloud, it comes with its own set of trade-offs too and that’s where Edge Computing steps in to offer the perfect complement to cloud computing.
Edge Computing places processing power closer to where data is being created and offers the following advantages over Cloud:
- Lower Latency: Since data is processed where it is created, there is no longer a latency to ship the data to the data center in order to process them.
- Lower bandwidth: When large amounts of data are being generated, it becomes inefficient and expensive to send all of this data to the data centers.
- Fault Tolerance: Cloud computing breaks down when the network is unreliable and systems have to function even when the network connection to cloud is down.
- Privacy: In many applications, sensitive data has to remain in-house and cannot be shipped to the cloud due to privacy regulations.
A related, but somewhat emerging field is “edge cloud computing”. In this paradigm, cloud providers are partnering with telecom companies to bring their platforms closer to the edge. By using 5G networks, they can bring the cloud technology to devices that were previously not accessible via a traditional WI-Fi connection. This is certainly exciting.
Zippin’s checkout-free solution takes advantage of the best of both the worlds. Zippin’s platform works through sensor fusion of shelves as well as camera sensors. We use state-of-the-art edge AI accelerators to process these data at the edge and send the processed stream to the cloud for further processing, thereby optimizing the compute required in stores as well as the bandwidth required for the cloud. Since the bulk of the data is processed at the edge, our solution offers low latency and remains privacy-friendly. We have built our solution to be fault tolerant; even when internet connection is down for a few hours, the store can be fully operational.
NVidia’s GTC 21 conference is just around the corner. I am looking forward to some exciting announcements and maybe even some surprises from Nvidia! GPUs and other AI specific integrated chips are the key drivers for applications of AI in autonomous vehicles, factory and home automation and retail transformations. These Application Specific Integrated Chips (ASIC’s) form the backbone of computing that scales from edge to the data center to the cloud.
Edge computing has seen tremendous growth in recent years due to widespread adoption of connected devices/sensors as well as cameras. Large companies such as Apple, IBM, Intel, Facebook, Amazon and Google are all designing their own AI ASICs/accelerators using lower-precision arithmetic to drive down power and heat generated. There are also a number of rapidly emerging startups in this space. It will be interesting to see what NVidia has to unveil at GTC this year and how that will enable more AI at the edge, as well as the cloud, while being extremely power efficient.