Shreyas Shah, Principal Data center Architect, Xilinx
In the cloud computing era, data growth is exponential. Every day billions of photos are shared and large amount of new data created in multiple formats. Within this cloud of data, the relevant data with real monetary value is small. To extract the valuable data, big data analytics frame works like SparK is used. This can run on top of a variety of file systems and data bases. To accelerate the SparK by 10-1000x, customers are creating solutions like log file accelerators, storage layer accelerators, MLLIB (One of the SparK library) accelerators, and SQL accelerators etc.
FPGAs (Field Programmable Gate Arrays) are the ideal fit for these type of accelerators where the workloads are constantly changing. For example, they can accelerate different algorithms on different data based on end users and the time of the day, but keep the same hardware.
This webinar will describe the role of FPGAs in SparK accelerators and give SparK accelerator use cases.