Manufacturers are turning to edge computing to develop real-time data insights. But companies are finding that an edge-to-cloud approach is often best for deep learning.
Manufacturers hoping to optimize their workflows are increasingly turning to artificial intelligence (AI) and deep learning to detect patterns and efficiencies that can't be spotted by the human eye. But integrating machine learning into the manufacturing process isn't as simple as installing a piece of software or hardware. It also requires new approaches to storage and data processing.
Because the amount of data produced for deep learning applications is now so voluminous, manufacturers need new systems to process it at the pace of business. Enter edge computing. By using edge computing to process data locally, organizations have been able to avoid the undesirable latencies that come with transmitting data to a cloud or centralized server for processing.
An edge-only approach does have its drawbacks. While edge computing can provide valuable real-time analysis of local data, living exclusively on the edge can deprive companies of the greater value that can come from aggregating edge-processed data in a central server. To take full advantage of AI and deep learning, manufacturers need both the rapid processing capabilities of edge networks and the ability to aggregate it all in a central environment.
Edge-to-cloud: More powerful than edge alone
Processing data at the edge while also enabling efficient migration to and from the cloud enables deep learning solutions to provide greater insight and power. That's why we at HPE have been helping manufacturers adopt an edge-to-cloud approach.
Seagate one of many companies we've helped take advantage of this strategy. The data storage company deploys HPE's edge computing and deep learning tools to improve quality assurance in its factories. This includes the HPE Edgeline EL4000 Converged Edge System, which empowers Seagate to collect around 15 million product images a day across all of its factories for deep learning.
Deep learning also helps Seagate detect patterns in factory data that would otherwise be missed by human analysts. This offers great benefits for improving the regulation of factory operations and the development of new strategies for enhanced workflows. But it wouldn't be possible with a purely edge-based approach because edge computing alone isn't capable of processing the vast quantities of data needed for deep learning.
This is where high-performance computing and the cloud come in. To get the most of the HPE edge computing and artificial intelligence tools it employs, Seagate also uses the HPE Apollo 6500 System. The system is effectively a supercomputer, and because it comes equipped with eight NVIDIA Tesla GPUs, it's capable of providing the data-crunching power needed to facilitate deep learning. It's the combination of the Apollo 6500 Gen10 system and the EL4000 that gives Seagate the insights it needs.
Predicted growth in edge computing
One of the many reasons why we believe edge-to-cloud computing to be so advantageous is because it doesn't necessarily require a radical change in computing infrastructure. Rather, it requires a change in how existing infrastructure and technologies are combined. Even as HPE's edge-to-cloud tools offer distinct operational gains, they don't necessarily require that an organization fundamentally alter its practices to implement them. Seagate found the integration of the Apollo 6500 and the EL4000 into its workflows seamless because both operated similarly to the existing servers in its data centers.
As industries realize that an edge-to-cloud approach to machine learning can yield actionable insights, we expect this sector to enjoy robust growth. BIS Research estimates that the global edge computing market will grow 22 percent every year, eventually reaching a net value of $13.44 billion by the end of 2025. Similarly, Markets and Markets foresees the global market for edge-based AI software rising from $356 million in 2018 to $1.15 billion in 2023—an annual growth rate of 26.5 percent.
Embracing an edge-to-cloud future
We're already seeing the promised value of edge computing being delivered. Seagate is using HPE's edge and AI tools to detect defects in its storage products as they're being manufactured. What's more, thanks to HPE's edge computing and deep learning tools, Seagate can predict if a product will be vulnerable to developing a defect later on. HPE's edge-to-cloud AI system learns to spot potential flaws by analyzing millions of images in real time, enabling Seagate to preemptively rectify affected products without having to schedule separate maintenance events or disrupt the manufacturing process.
It's hard to calculate just how much time and money this system saves for Seagate, but time and cost savings are why manufacturers are turning to edge-to-cloud solutions for machine learning. Indeed, edge-to-cloud frameworks are likely to innovate across myriad industries—to the point where some organizations are even proposing creating smarter cities with it. In this environment, manufacturers who don't adopt an edge-to-cloud approach might end up missing out on a more efficient and productive future.
Watch and learn more about how Seagate optimizes manufacturing using HPE's Converged Edge Systems and AI analytics.
Pankaj Goyal Vice President, HPE AI Business Hewlett Packard Enterprise
Pankaj is building HPE’s Artificial Intelligence business. He is excited by the potential of AI to improve our lives, and believes HPE has a huge role to play. In his past life, he has been a computer science engineer, an entrepreneur, and a strategy consultant. Reach out to him to discuss everything AI @HPE.