Loading...
Using FPGA as Accelerator for Processing Units in Big Data Stream Processing Engine
Darjani, Armin | 2017
604
Viewed
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 50190 (19)
- University: Sharif University of Technology
- Department: Computer Engineering
- Advisor(s): Goudarzi, Maziar
- Abstract:
- Distributed stream processing frameworks (DSPFs) are used for real-time processing of big data. Apache Storm is one of the most popular stream processing systems in industry today. By increasing data generation rate we need new methods to overcome processing requirements of DSPFs like Apache Storm. In this thesis we investigate the feasibility of incorporation FPGA acceleration into Apache Storm. Using FPGAs as co-processors in powerful servers can improve performance and accelerate processing of streaming data by increasing parallelism, decreasing processing time of each processing units and decreasing communication delay between these units. Our design includes a hardware part that contains controlling modules and interfaces to create user ip cores and one software section that contains driver, user library and software architecture for proper use of hardware section. Also a scheduling algorithm is proposed for proper use of our design in scale. This algorithm should guess number of each executive components and how they execute(SW only or HW/SW) for maximizing streaming applications throughput. In addition to decreasing communication overheads of using FPGAs as accelerators, we tried not to change Apache Storm framework to preserve its extensibility and fault tolerance attributes. The experimental results shows that our design can decrease end to end latency of data processing significantly and achieve up to 35x throughput gain in some of benchmarks
- Keywords:
- Accelerators ; Big Data ; Throughput ; Data Center ; Big Data Proccessing ; Field Programmable Gate Array (FPGA) ; Apache Storm ; Stream Data Processing
- محتواي کتاب
- view