Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation

Miryeong Kwon; Donghyun Gouk; Eunjee Na; Jiseon Kim; Junhee Kim; Hyein Woo; Eojin Ryu; Hyunkyu Choi; Jinwoo Baek; Hanyeoreum Bae; Mahmut Kandemir; Myoungsoo Jung

doi:10.1109/MM.2025.3574261

Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation

Miryeong Kwon, Donghyun Gouk, Eunjee Na, Jiseon Kim, Junhee Kim, Hyein Woo, Eojin Ryu, Hyunkyu Choi, Jinwoo Baek, Hanyeoreum Bae, Mahmut Kandemir, Myoungsoo Jung

Research output: Contribution to journal › Article › peer-review

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 2.0× better performance for I/O-intensive workloads, and 7.9× improvement in distributed LLM inference.

Original language	English (US)
Journal	IEEE Micro
DOIs	https://doi.org/10.1109/MM.2025.3574261
State	Accepted/In press - 2025

All Science Journal Classification (ASJC) codes

Software
Hardware and Architecture
Electrical and Electronic Engineering

Access to Document

10.1109/MM.2025.3574261

Cite this

@article{00b238c0e9d449c9a651eb3c393c2304,

title = "Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation",

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 2.0× better performance for I/O-intensive workloads, and 7.9× improvement in distributed LLM inference.",

author = "Miryeong Kwon and Donghyun Gouk and Eunjee Na and Jiseon Kim and Junhee Kim and Hyein Woo and Eojin Ryu and Hyunkyu Choi and Jinwoo Baek and Hanyeoreum Bae and Mahmut Kandemir and Myoungsoo Jung",

note = "Publisher Copyright: {\textcopyright} 1981-2012 IEEE.",

year = "2025",

doi = "10.1109/MM.2025.3574261",

language = "English (US)",

journal = "IEEE Micro",

issn = "0272-1732",

publisher = "IEEE Computer Society",

}

TY - JOUR

T1 - Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation

AU - Kwon, Miryeong

AU - Gouk, Donghyun

AU - Na, Eunjee

AU - Kim, Jiseon

AU - Kim, Junhee

AU - Woo, Hyein

AU - Ryu, Eojin

AU - Choi, Hyunkyu

AU - Baek, Jinwoo

AU - Bae, Hanyeoreum

AU - Kandemir, Mahmut

AU - Jung, Myoungsoo

PY - 2025

Y1 - 2025

N2 - 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 2.0× better performance for I/O-intensive workloads, and 7.9× improvement in distributed LLM inference.

AB - 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 2.0× better performance for I/O-intensive workloads, and 7.9× improvement in distributed LLM inference.

UR - https://www.scopus.com/pages/publications/105007759689

UR - https://www.scopus.com/inward/citedby.url?scp=105007759689&partnerID=8YFLogxK

U2 - 10.1109/MM.2025.3574261

DO - 10.1109/MM.2025.3574261

M3 - Article

AN - SCOPUS:105007759689

SN - 0272-1732

JO - IEEE Micro

JF - IEEE Micro

ER -

Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation

Abstract

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this