Data-protection class workloads, including backup and long-term retention of data, have seen a strong industry shift from tape-based platforms to disk-based systems. But the latter are traditionally designed to serve as primary storage and there has been little published analysis of the characteristics of backup workloads as they relate to the design of disk-based systems. In this paper, we present a comprehensive characterization of backup workloads by analyzing statistics and content metadata collected from a large set of EMC Data Domain backup systems in production use. This analysis is both broad (encompassing statistics from over 10,000 systems) and deep (using detailed metadata traces from several production systems storing almost 700TB of backup data). We compare these systems to a detailed study of Microsoft primary storage systems [22], showing that backup storage differs significantly from their primary storage workload in the amount of data churn and capacity requirements as well as the amount of redundancy within the data. These properties bring unique challenges and opportunities when designing a disk-based filesystem for backup workloads, which we explore in more detail using the metadata traces. In particular, the need to handle high churn while leveraging high data redundancy is considered by looking at deduplication unit size and caching efficiency. 1

Abstract—Key to the compression-capability of a data deduplication system is the definition of redundancy. Traditionally, two data items are considered redundant if their underlying bit-streams are identical. However, this notion of redundancy is too strict for many applications. For example, for a video storage platform, two videos encoded in different formats would be unique at the system level but redundant at the content level. Intuitively, introducing application-level intelligence in redundancy detection can yield improved data compression. We propose ViDeDup (Video De-Duplication), a novel framework for video de-duplication based on an application-level view of redundancy. The framework goes beyond duplicate data detection to similarity-detection, thereby providing application-level knobs for defining acceptable level of noise during replica detection. Our results show that by trading CPU for storage, a 45 % reduction in storage space could be achieved, in comparison to 8 % yielded by system level de-duplication for a dataset collected from video sharing sites on the Web. We also present tradeoff analysis for various tunable parameters of the system to optimally tune the system for performance, compression and quality. I.