Date of Original Version

7-2008

Type

Technical Report

Rights Management

All Rights Reserved

Abstract or Description

High-performance parallel file systems are a critical component of the largest computer systems, are primarily proprietary, and are specialized to high end computing systems that have many access patterns known to be unusual in enterprise and productivity workplaces. Yet little knowledge of even the basic distributions of file systems and file ages are publicly available, even though significant effort and importance is increasingly associated with small files, for example. In this paper we report on the statistics of supercomputing file systems at rest from a variety of national resource computing sites, contrast these to studies of the 80s and 90s of academic and software development campuses and observe the most interesting characteristics in this novel data.

Share

COinS