Dell PowerEdge customers wanting detailed Hadoop tuning information, look no further! TheTuning Hadoop on Dell PowerEdge Server whitepaper has just been released, and contains a wealth of information to allow the user to optimize their Hadoop workloads. Included is a guide for optimization, and the performance effects of various hardware, firmware and Hadoop configurations options are discussed and charted in great detail.
The Four primary areas covered in detail are:
- Hardware Selection
- BIOS and Firmware options
- OS and Filesystem options
- Hadoop Configuration options
Key findings in this whitepaper are the following:
- Compression increases performance by up to 20 percent under certain Hadoop activities.
- Block size can increase Hadoop MapReduce performance by up to 21 percent.
- The number of hard disk drives (HDD) attached to the data nodes can decrease the Hadoop job execution time by up to 20 percent for each HDD added, until storage-bus saturation is achieved.
- Changing certain file system settings in the native OS increased Hadoop performance by up to 10 percent.
- Changing certain default OS settings can increase Hadoop performance by up to 15 percent and increase the chance of completion of Hadoop workloads.
- By adding certain Java options to the Hadoop JVMs, Hadoop performance can increase by up to five percent.
- By changing certain Hadoop parameters, performance can increase by up to 12 percent in MapReduce jobs.
For much more detailed information, please consult the whitepaper.
Image may be NSFW.Clik here to view.
