Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kwon, Miryeong, Gouk, Donghyun, Na, Eunjee, Kim, Jiseon, Kim, Junhee, Woo, Hyein, Ryu, Eojin, Choi, Hyunkyu, Baek, Jinwoo, Bae, Hanyeoreum, Kandemir, Mahmut, Jung, Myoungsoo
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913884705652736
author Kwon, Miryeong
Gouk, Donghyun
Na, Eunjee
Kim, Jiseon
Kim, Junhee
Woo, Hyein
Ryu, Eojin
Choi, Hyunkyu
Baek, Jinwoo
Bae, Hanyeoreum
Kandemir, Mahmut
Jung, Myoungsoo
author_facet Kwon, Miryeong
Gouk, Donghyun
Na, Eunjee
Kim, Jiseon
Kim, Junhee
Woo, Hyein
Ryu, Eojin
Choi, Hyunkyu
Baek, Jinwoo
Bae, Hanyeoreum
Kandemir, Mahmut
Jung, Myoungsoo
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2506_06769
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation
Kwon, Miryeong
Gouk, Donghyun
Na, Eunjee
Kim, Jiseon
Kim, Junhee
Woo, Hyein
Ryu, Eojin
Choi, Hyunkyu
Baek, Jinwoo
Bae, Hanyeoreum
Kandemir, Mahmut
Jung, Myoungsoo
Hardware Architecture
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.
title Containerized In-Storage Processing and Computing-Enabled SSD Disaggregation
topic Hardware Architecture
url https://arxiv.org/abs/2506.06769