Home › Categories › Technology› MapReduce Evaluation

MapReduce Evaluation.pdf

kkc15a.pdf

🔗 Source: freeh.wordpress.ncsu.edu

📊 Size: 195 KB

📄 Pages: 8 pages

⬇️ Downloads: 131

Summary

MapReduce applications on a large cluster of 540 nodes were evaluated, focusing on scaling behavior, parallelism and overlap configuration, and general observations. Results showed that IO-intensive applications do not scale with increasing data size, and that different applications require different levels of parallelism and overlap to minimize completion time. Additionally, the importance of appropriate memory allocation and reducing log file size was highlighted. The study found that applications with high IO requirements have completion times that increase disproportionately with data size, while those with medium to low IO requirements have completion times that increase proportionally. An appropriate configuration of parallelism and overlap was also found to improve application performance, with the optimal configuration depending on the application's CPU and IO characteristics.

Download File

Description

MapReduce applications on a large cluster of 540 nodes were evaluated, focusing on scaling behavior, parallelism and overlap configuration, and general...

Technical Information

File Format: PDF
File Size: 195 KB
Pages: 8
Language: EN
Total Downloads: 131
Last Updated: 2 hours ago

Document Overview

This PDF document about MapReduce Evaluation provides comprehensive information and guidance. Whether you're a beginner or advanced user, this resource offers valuable insights into MapReduce Evaluation.