?

FlexPod Outperforms JBOD for Hadoop at Scale

This white paper explores the FlexPod Select architecture for Hadoop versus JBOD (“Just a Bunch of Disks”) architecture from both a technical and financial perspective.

Executive Summary

Is commodity or white box infrastructure a viable platform for enterprise organizations? As with most technical decisions, the answer to that question is “it depends.” Use cases are important. This paper will explore the FlexPod Select architecture for Hadoop versus JBOD (“Just a Bunch of Disks”) architecture from both a technical and financial perspective.

JBOD or white box, commodity infrastructure often represents a low barrier to entry from a cost standpoint and can be a prudent decision when uncertainty exists with regard to the success of the workloads that will be placed on the environment. Our testing investigated the effect of data disk failure on network and compute capabilities within a “like-to-like” environment.

A critical assumption for this analysis is that the use cases have been identified and the architecture is supporting business critical and valuable workloads requiring scale across an enterprise organization. These workloads do not represent a science experiment; value is proven and scale is paramount. In this instance the FlexPod Select is a clear choice on technical and financial merits over JBOD architecture.

Under the aforementioned critical assumption, the following key takeaways result and can be attributed to the FlexPod Select architecture for Hadoop:

  • Businesses can invest in Human Resources (HR) rather than IT infrastructure
  • HR investments can be redirected from support technicians to data scientists
  • Data scientists deliver business value through real-time analytics, trends and insights driving: revenue growth, cost control, improved productivity, risk mitigation and customer retention