Machine- and user-generated data and content provide a massive opportunity for the modern enterprise, delivering business insights that help to optimize operations, develop products that delight customers and facilitate positive change in our society. But harnessing the power of Big Data is not easy – the sheer volume of data creates unprecedented challenges around secure storage, retrieval, processing and analysis.
While traditional storage and server architectures rely on moving all data through a host CPU for processing, this creates a bottleneck between storage and the CPU that can limit access to processing resources, create substantial delays and result in unreliable performance. These problems will only intensify as the amount of data generated by businesses continues to skyrocket. IDC predicts new data creation will grow at a CAGR of 23%, reaching 175 zettabytes by 2025↗
Samsung is committed to developing extremely high-performance, energy-efficient computational storage solutions to power an array of data-intensive applications and use cases, from AI and machine learning to analytics, virtualization, 5G and others. With the recent announcement
of its second-generation SmartSSD® computational storage drive (CSD), we have transcended the boundaries of what is possible in computational storage.
Samsung + AMD: Turning Big Data into Fast Data
Powered by the Xilinx Adaptive Platform, Samsung’s second-generation SmartSSD is the ideal customizable, programmable computational storage platform. It performs high-speed computations of big data wherever it is stored and offers a flexible platform for developers to innovate groundbreaking solutions.
How does it accomplish these performance improvements? By processing data directly, where it resides. By integrating a powerful and flexible Xilinx Versal™ Adaptive SoC from AMD into a solid-state drive (SSD) from Samsung, computation can occur in parallel at the drive level. This minimizes data transfers between the CPU, GPU and RAM, reducing the energy spent on moving data and allowing the host CPU to handle higher-level tasks more efficiently. Compared to conventional data center SSDs, processing time for scan-heavy database queries can be slashed by over 50%, energy consumption by up to 70% and CPU utilization by up to 97%.