Arno

Arno (Adjustable Resource-NAND Orchestrator) is a scalable SSD simulator implemented in Python 3.11.
It allows performance and latency analysis based on customizable NAND configurations and supports simulation under various I/O traces and workload scenarios.
Additionally, Arno provides QoS evaluation capabilities, making it suitable for studying real-world SSD behavior under different conditions.

---

Features

• Performance Metrics: Measure read/write throughput and latency
• Workload Patterns: Supports sequential and random read/write
• Customizable Configuration: Easily change number of channels, ways, and planes
• Predefined Workload:
  • performance: Includes 4 basic scenarios (sequential/random read/write)

---

Requirements

• Python 3.11
• Required packages are listed in requirements.txt

---

How to Run

To execute the simulator:

python storage_simulator.py --pre-defined-workload performance

To view available options:

python storage_simulator.py --help

---

Configuration Options

Argument	Type	Default	Description
--pcie-gen	int	5	PCIe generation number. Choose from {4, 5, 6}
--channel	int	16	Number of NAND channels
--way	int	2	Number of ways per channel
--plane	int	4	Number of planes per die
--nand-io-mbps	int	0	NAND interface bandwidth in Mbps
--nand-product	str	'TLC_EXAMPLE'	NAND model (more types to be supported later)
--nand-cell-type	str	TLC	NAND type (Supports SLC, MLC, TLC)
--gc-threshold	float	0.2	The threshold ratio for triggering GC
--urgent-gc-threshold	float	0.05	The threshold ratio for triggering urgent GC (Urgent GC halts host write)
--sustained-block-rate	float	0.8	Ratio of blocks to fill under sustained conditions
--sustained-page-rate	float	0.8	Ratio of pages to fill under sustained conditions
--make-sustained	flag	True	Make NAND into sustained state
--workload-type	str	'basic'	Workload type option ('pcmark10' is only supported.)
--pre-defined-workload	str	None	Select a predefined workload scenario
--range-bytes	str	None	Total data range (Bytes) of each TestCase
--folder-name	str	None	Name of the output folder for results
--enable-command-record	flag	False	Enable recording of command traces
--enable-performance-record	flag	False	Enable recording of performance logs
--enable-utilization	flag	False	Enable recording of resource utilization

Example

# Use a pre-defined workload (any --range-bytes will be ignored)
python storage_simulator.py --pre-defined-workload performance_256MB

# Use the default workload 'performance' with a custom size
python storage_simulator.py --range-bytes 512MB

# Run with all defaults (default workload 'performance' and its default size)
python storage_simulator.py

# Run with pcmark10 I/O trace
python storage_simulator.py --workload-type pcmark10

# Other example
python storage_simulator.py --pre-defined-workload performance_512MB \\  
  --pcie-gen 5 \\
  --channel 16 --way 2 --plane 4 \\
  --nand-product TLC_EXAMPLE \\
  --enable-performance-record --folder-name results/test

---

Output

• Average latency
• Total number of I/O operations
• Throughput (MB/s)
• Latency distribution and statistics

---

Project Structure

Arno/
├── core/                   # Core logic and behavior (NAND, Buffer, Bus, etc.)
├── product/general/        # SSD-level functionality implementation
├── output/                 # Stores simulation results
├── storage_simulator.py    # Main entry point for simulation
├── requirements.txt
└── README.md

License

This project is distributed under the terms of the MIT License (see LICENSE).

It also includes third-party libraries that are subject to their own license terms. For details, please refer to THIRD_PARTY_LICENSES.