As Yuan Yuan, President of Huawei Scale-Out Storage Domain, recently wrote, all-flash data storage is the future—and for good reason. She thinks it will accelerate important progress in artificial intelligence (AI), too. Yuan adds that in a world of rapidly increasing data demand, he doesn’t have any intention of slowing down AI capabilities. For instance, his capabilities won’t slow down, he says. Providing the technological underpinning to AI development, he points out, is data, making advanced storage and analytics tools increasingly critical to shaping its future.
Today, the avg CPU card utilization rate is just 50%, showing there is plenty of opportunity to be more efficient. Additionally, Yuan claims that Huawei’s all flash storage systems help customers improve performance by up to 54 times, while cutting energy use and floor space by 90%. By leveraging these innovative solutions, enterprises can proactively meet their demand for more data without the added expense of more power and space.
Efficiency and Reliability of Flash Storage
Yuan points out the major breakthroughs on the efficiency front that flash storage technology has made. These advancements make it distinct from other older hard disk drives (HDD). Flash storage uses approximately an order of magnitude less power than traditional storage architectures. Moreover, it’s a lot lighter and more compact. Beyond the beauty of this new technology, this transition isn’t just technological—it’s a key step toward more sustainable data stewardship.
He points out the reliability of today’s solid-state drives (SSDs). Due to improvements in SSD controller technology and NAND flash memory components, these drives now provide lifespans comparable to older HDD technologies, rated at approximately two million hours. Yuan affirms that current SSDs can last five to ten years without issues, enhancing their appeal for businesses looking for dependable storage solutions.
“Data is the future. AI needs data, and storage is the foundation that will drive new productivity.” – Yuan Yuan
Yuan adds that Huawei’s all-flash storage systems have done particularly well in energy efficiency. They have very low operating costs, using only 0.25 watts per terabyte, more than double the industry average. This new level of efficiency is much better than conventional storage, which generally uses 0.5 to 1 watt per terabyte. Beyond its usual notches, Huawei includes lots of little touches to reduce power consumption, like automatically lowering CPU speeds when workloads are light.
Addressing the Growing Data Demands
The accelerating pace of data creation is an ever-present reality for every sector. Yuan explains that AI models require enormous datasets to ever improve their performance. He adds that conventional HDDs are no longer sufficient for AI/big data analytics/high-concurrency applications that call for lower latency and greater input/output operations per second (IOPS).
He outlines how Huawei’s approach enables customers to double their storage capacity every two years without needing extra power or additional space. This forward-thinking approach not only helps companies stay ahead of their growing data needs, but empowers them to operate with more nimble infrastructure.
“The performance of traditional HDDs can no longer meet the demands of AI, big data, and high-concurrency applications for low latency and high IOPS.” – Yuan Yuan
According to Yuan, with Huawei’s all-flash storage solutions bottlenecks are completely removed, allowing the business to operate more efficiently and effectively. Recently, the company commercially released the Huawei OceanStor Pacific 9928 and 9926. These all-flash, scale-out storage systems are specifically designed for performance-intensive, big data use cases. Optimized for new file content, these systems use intelligent auto-tiering that increases performance. They address major drawbacks associated with legacy HDDs, such as power consumption.
The Shift Towards a New Storage Paradigm
Flash storage certainly provides a number benefits. It’s powerful technology, though at a premium price. At the moment, it’s approximately five times as expensive as regular HDDs. Yuan feels that this once insurmountable landscape is shifting quickly. With AI applications continuing to emerge and capture headlines, customers—both enterprise and consumer—will demand proven, logical, fast, and powerful storage solutions. This requirement will drive them to start using flash.
“It’s currently about three times more expensive than hard drives but things are changing quickly. AI needs huge amounts of data, and we now have better ways to manage and use that data.” – Yuan Yuan
Huawei’s continued investment in innovation will see the development of GPU+NPU native disk enclosures to further improve performance and efficiency. The company has dozens of other features in active development, including power saving and data compression. Their focus is on easing and lowering the cost of flash storage adoption for their customers.
Beyond that, Yuan is sensitive to the challenges of directing different kinds of data. To address this challenge, Huawei supports their clients to create their own local databases. Maintaining up-to-date data. These federal databases help AI models stay current and improve AI model performance. He acknowledges that while managing different types of data can be complicated, optimizing storage solutions is essential for organizations aiming to leverage AI effectively.
“For AI inference, we’ve added two key features. First, we use a unified cache manager (UCM) to store important data, so the system doesn’t have to recalculate information every time; instead, it can quickly access what it needs.” – Yuan Yuan