FSSD-EM: FPGA-based SSD emulator with energy modeling

Yu, Luyang

Permalink

https://hdl.handle.net/2142/120173 Copy

Description

Title

FSSD-EM: FPGA-based SSD emulator with energy modeling

Author(s)

Yu, Luyang

Issue Date

2023-05-04

Director of Research (if dissertation) or Advisor (if thesis)

Chen, Deming

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• SSD Emulation
• FPGA

Abstract

Solid State Drives (SSDs) have gained significant popularity due to their superior access latency, bandwidth, and energy efficiency compared to Hard Disk Drives (HDDs). To fully understand the impact of SSD design and microarchitecture on end-to-end application performance, researchers must delve deeper than treating SSDs as black-box components. Unfortunately, acquiring multiple SSDs for research is costly and inefficient, as the underlying microarchitecture remains hidden from the system designer. Although simulators have become the preferred method for studying SSDs, existing software-based simulators are limited in terms of real data transfers and cannot simulate the overhead associated with NVMe and PCIe interfaces. Additionally, simulating an entire SSD using software is a time-consuming process, constraining the number of experiments that can be conducted within a reasonable timeframe. To address these challenges, we introduce FSSD, an FPGA-based emulation system that accurately models NAND flash access patterns, the latency, and the energy consumption of a real NVMe SSD. FSSD capitalizes on the flexibility of an FPGA, enabling users to customize SSD microarchitecture features and explore the design space for data-intensive applications. Based on FSSD, we further introduce FSSD-EM with a built-in Garbage Collection (GC) algorithm and energy modeling module. In contrast to most software simulators that rely on Virtual Machines (VM), FSSD-EM is capable of interacting with real Operating Systems (OS). Our evaluations reveal that FSSD-EM provides a remarkable 1000x speedup compared to software-based simulation using SimpleSSD. FSSD-EM's ability to adjust SSD parameters and emulate NAND latency and energy cost with high precision makes it a valuable tool for SSD research and development efforts.

Graduation Semester

2023-05

Type of Resource

Thesis

Copyright and License Information

Copyright 2023 Luyang Yu

Owning Collections