Arash Tavakkol, Juan Gómez-Luna, and Mohammad Sadrosadati, ETH Zürich; Saugata Ghose, Carnegie Mellon University; Onur Mutlu, ETH Zürich and Carnegie Mellon University
Solid-state drives (SSDs) are used in a wide array of computer systems today, including in datacenters and enterprise servers. As the I/O demands of these systems have increased, manufacturers have evolved SSD design to keep up with this demand. For example, manufacturers have introduced new high-bandwidth interfaces to replace the traditional SATA protocol. These new interfaces, such as the NVMe protocol, are designed specifically to enable the high amount of concurrent I/O that SSDs are capable of delivering.
While modern SSDs with sophisticated features such as the NVMe protocol are already on the market, SSD simulation tools have fallen behind, as they do not capture these new features. We compare the outputs of these simulators to the performance measured from real off-the-shelf SSDs, and find three major shortcomings in state-of-the-art SSD simulators. First, existing simulators do not model critical features of new protocols like NVMe, such as their use of multiple application-level queues for requests and the elimination of OS intervention. Second, existing simulators do not capture the effects of advanced SSD maintenance algorithms (e.g., garbage collection), as they do not properly emulate the steady-state conditions that exist in real SSDs. Third, existing simulators do not capture the full end-to-end latency of I/O requests, which can incorrectly skew the simulated behavior of SSDs that use emerging memory technologies, such as 3D XPoint. We show that without the accurate modeling of these features, results from existing simulators deviate significantly from real SSD performance.
In this work, we introduce a new simulator, called MQSim, that accurately models the performance of both modern SSDs and conventional SATA-based SSDs. MQSim faithfully models new high-bandwidth protocol implementations, steady-state SSD conditions, and the full end-to-end latency for requests in modern SSDs. We validate MQSim using several modern SSDs, and show that MQSim uncovers several real and important issues that were not captured by existing simulators, such as the performance impact of inter-flow interference in modern SSDs. We plan to release MQSim as an open-source tool, and we hope that it can enable several new research directions in the future.
Open Access Media
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.
author = {Arash Tavakkol and Juan G{\'o}mez-Luna and Mohammad Sadrosadati and Saugata Ghose and Onur Mutlu},
title = {{MQSim}: A Framework for Enabling Realistic Studies of Modern {Multi-Queue} {SSD} Devices},
booktitle = {16th USENIX Conference on File and Storage Technologies (FAST 18)},
year = {2018},
isbn = {978-1-931971-42-3},
address = {Oakland, CA},
pages = {49--66},
url = {https://www.usenix.org/conference/fast18/presentation/tavakkol},
publisher = {USENIX Association},
month = feb
}