It’s amazing to watch our customers evolving through their Big Data journey. Our conversations have changed over the last three years as the workload matures and trickles into mainstream IT. This is great for customers as it enables them to get smart about their data and reap the benefits from those insights at a lower cost compared to traditional BI solutions.
A couple of years ago, we were hearing, “What is Hadoop and what can it do for my business?” Six to nine months later, the question became, “What does the architecture look like, how do I get data in/out, and how do I deploy a proof-of-concept?” Today, customers are asking, “How can I get more work and faster results out of my Hadoop cluster?”
Based on this need, Dell wanted to provide some insights into how a Hadoop cluster, when properly tuned, can result in better performance. We enlisted a team of Dell engineers to gather Hi-bench 2.2 benchmark data to better understand the capabilities and performance of the Dell | Cloudera Reference Architecture we built. The goal of the study was to tune and optimize the performance of the cluster plus identify and clear bottlenecks.
Through in-depth analysis and benchmark expertise, Dell Engineering was able to improve performance by 50% on a 12 node cluster compared to an un-tuned cluster running the same configuration and Hadoop distribution. The engineering team was able to pinpoint tuning and refine adjustments to O/S parameters, HDFS block size, and Hadoop configuration settings within the Cloudera Distribution of Hadoop (CDH) allowing greater performance for MapReduce.
To read the full performance report so you can learn more about how to better utilize and get more work out of your Hadoop cluster click here or contact a sales specialist for more details.
Dell is all about helping customers make the right decisions whether it’s architecture, deployment tools, or configuration. We’re dedicated to providing the best set of tools so customers can get the most out of their investment. We are with you every step of the way.