Measuring Performance of Solid State Arrays – Tech Insight Paper is intended for use by our End-User clients to understand the technology behind the product, and the impact that it has on the industry. Our Tech Insight Series provides insight into current and emerging IT trends, providing topical and product discussion for managers and executives.
The advent of datacenter consolidation with server and desktop virtualization has created highly demanding and randomized, latency-sensitive workloads that rival the demands of mission critical database environments. Solid state storage (SSS) arrays are an increasingly important solution within modern IT infrastructures to accelerate specifically Tier-1 applications, virtual infrastructures, and the next generation of computing with modern applications, such as Hadoop and powerful analytics applications.
The performance of a storage array with built-in data reduction technologies is dependent upon the breadth of the data content, metadata characteristics, and access patterns presented to an array. Today, most data sets and data streams produced by current I/O generation tools to measure data reduction storage arrays are inadequate to demonstrate actual performance levels. But that is changing and there are a small number I/O generation tools that have added the appropriate level of complexity to measure the performance of modern SSS arrays. This document first presents an approach to the steps needed to create appropriate data sets and data streams and then how to measure performance once they are presented to the storage array.
The performance methodology developed in this document addresses the specific requirements to measure the performance of enterprise SSS arrays using advanced capacity optimization technologies. The SNIA has done a great job in publishing a specification to measure raw performance for individual solid state devices (SSD), but a way is still needed to accurately measure SSS arrays at a system level supporting advanced inline data services, like data reduction technologies.
Other posts and documents:
Plan subscribers may also be interested in: