Simulation of Hierarchical Storage Systems for TCO and QoS

Abstract

Due to the variety of storage technologies deep storage hierarchies turn out to be the most feasible choice to meet performance and cost requirements when handling vast amounts of data. Long-term archives employed by scientific users are mainly reliant on tape storage, as it remains the most cost-efficient option. Archival systems are often loosely integrated into the HPC storage infrastructure. In expectation of exascale systems and in situ analysis also burst buffers will require integration with the archive. Exploring new strategies and developing open software for tape systems is a hurdle due to the lack of affordable storage silos and availability outside of large organizations and due to increased wariness requirements when dealing with ultra-durable data. Lessening these problems by providing virtual storage silos should enable community-driven innovation and enable site operators to add features where they see fit while being able to verify strategies before deploying on production systems. Different models for the individual components in tape systems are developed. The models are then implemented in a prototype simulation using discrete event simulation. The work shows that the simulations can be used to approximate the behavior of tape systems deployed in the real world and to conduct experiments without requiring a physical tape system.

Cost and Performance Modeling for Earth System Data Management and Beyond

“…These efforts are mostly focused on finding new architectures to cope with exascale workloads and do not consider initial and operation cost specifically. In section Section 4 we also look at compatibility to a fine grained approach [13] which uses DES to determine cost and quality of service for hierarchical storage system including tape system in data centers. For the purpose of standardization and communication with users these models can be too fine-grained.…”

“…For additional insight, parts of a coarse grained model can be refined with more detailed models. For example, by mimicing workload execution using DES and workload traces in combination with the actual data center topology, taking temporal behavior into account (see Section 2 and [13]). This way one can narrow down on uncertain areas in the general model.…”

“…As described in Section 1 research institutions tend to organize compute and storage capabilities either by specializing at different locations or by collocating and consolidating different services into a single site, as is the case with DKRZ. In this paper only the summary of a larger case study is presented, a more detailed report [11] is available separately.…”