Скачать или смотреть Achieving Flexible, High Performance Data Lakes with Xilinx, Samsung SmartSSD®, and Bigstream

Achieving Flexible, High-Performance Data Lakes with Xilinx, Samsung SmartSSD®, and Bigstream Hyperacceleration Technology

Webinar - June 17, 2020

Summary:
In the last 10 years, data lakes have become the prominent methodology for storing and providing access to data for several data-driven enterprises. Data scientists, data analysts, and data engineer roles all make use of and/or support data lakes, albeit in a variety of unique ways. That said, the commonality across most data users is an ever-increasing demand for larger data, more data sources, and more dynamic data, making data lake performance paramount. Due to cost and physical constraints, CPU scale-up and scale-out are now ceasing to meet these performance demands. Advanced acceleration hardware is becoming readily available to help with performance, but the programming of these is very complex and ends up being a burden to the data analytics professional.

Computational storage offers a promising and viable alternative to address these performance needs. This approach combines advanced processing hardware along with state-of-the-art storage technology on a single platform. As such, it offers three key benefits: higher performance for processing data, offload of this computation which frees CPU resources, and higher effective server-storage connection bandwidth. Samsung’s SmartSSD® Computational Storage Drive (CSD) is the state-of-the-art in this genre, combining a powerful Xilinx FPGA card with high-performance storage to achieve all three of these benefits. In combination with Bigstream Hyperacceleration technology, the solution can solve both the performance and programmability problems for the end-user of data lakes.

In this webinar, we will demonstrate how the SmartSSD Computational Storage Drive can support the performance goals of a data lake, leveraging Bigstream’s Hyperacceleration technology for seamless performance improvements. Hyperacceleration automatically compiles analytics workloads onto the SmartSSD hardware, with zero code changes or additional developments to achieve performance results of 3-5x acceleration for key workloads to address a wide variety of performance issues while increasing productivity for all end-user types.


Speakers:
Bishwa Roop Ganguly
Chief Solutions Architect, Bigstream
Bishwa Roop Ganguly is Chief Solutions Architect at Bigstream Solutions. He has a PhD in Electrical Engineering from MIT, and an MS and BS in Computer Science from University of Illinois and University of California at Berkeley, respectively. He has published extensively in the field of parallel processing and computer networks. He also has 5 years of experience as a Data Scientist using Hadoop, Spark and SQL. He currently manages the customer engagements at Bigstream.

Dan Eaton
WW Business Development Lead, Database and Data Analytics, Xilinx
Dan Eaton is WW Business Development Lead, Database and Data Analytics at Xilinx. He is responsible for developing the company’s FPGA IaaS ecosystem with Amazon Web Services, onboarding hundreds of partners and thousands of developers. Dan joined Xilinx in 2017 and has been evangelizing FPGA accelerated computing solutions to the enterprise and public sector since 2013.

#hyperacceleration