Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation

2025-06-07

IEEE Micro

AI Summary

To address the high data migration overhead and underutilized SSD computational capability in storage-disaggregated architectures, this paper proposes a firmware-level lightweight containerized in-storage processing (ISP) framework. Our approach introduces two novel mechanisms: Ethernet-over-NVMe communication and a Virtual Firmware architecture, enabling native OS-level virtualization container support (e.g., Docker) directly within SSD firmware—thereby achieving secure and efficient storage-side compute offloading. By co-designing NVMe protocol extensions and storage-compute scheduling, our framework delivers a 2.0× speedup for I/O-intensive workloads and accelerates distributed large language model inference by 7.9×, while significantly reducing host CPU and memory overhead. This work establishes a new hardware-level co-design paradigm for disaggregated storage systems and AI workloads.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Distributed Search
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• Security and Privacy: Data transparency and provenance
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

ISP minimizes data transfer for analytics but faces challenges in adaptation and disaggregation. We propose DockerSSD, an ISP model leveraging OS-level virtualization and lightweight firmware to enable containerized data processing directly on SSDs. Key features include Ethernet over NVMe for network-based ISP management and Virtual Firmware for secure, efficient container execution. DockerSSD supports disaggregated storage pools, reducing host overhead and enhancing large-scale services like LLM inference. It achieves up to 2.0x better performance for I/O-intensive workloads, and 7.9x improvement in distributed LLM inference.

Problem

• Enables containerized data processing directly on SSDs
• Reduces host overhead in disaggregated storage pools
• Improves performance for I/O-intensive and distributed workloads

Innovation

• Containerized data processing on SSDs
• Ethernet over NVMe for ISP management
• Virtual Firmware for secure container execution