Delilah: eBPF-offload on Computational Storage

Niclas Hedam; Morten Tychsen Clausen; Philippe Bonnet; Sangjin Lee; Ken Friis Larsen

doi:10.1145/3592980.3595319

Delilah: eBPF-offload on Computational Storage

Niclas Hedam, Morten Tychsen Clausen, Philippe Bonnet, Sangjin Lee, Ken Friis Larsen

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

The idea of pushing computation to storage devices has been explored for decades, without widespread adoption so far. The definition of Computational Programs namespaces in NVMe (TP 4091) might be a breakthrough. The proposal defines device-specific programs, that are installed statically, and downloadable programs, offloaded from a host at run-time using eBPF. In this paper, we present the design and implementation of Delilah, the first public description of an actual computational storage device supporting eBPF-based code offload. We conduct experiments to evaluate the overhead of eBPF function execution in Delilah, and to explore design options. This study constitutes a baseline for future work.

Original language	English
Title of host publication	International Workshop on Data Management on New Hardware (DaMoN)
Place of Publication	Seattle, Washington, USA
Publication date	19 Jun 2023
DOIs	https://doi.org/10.1145/3592980.3595319
Publication status	Published - 19 Jun 2023
Event	Conference: International Workshop on Data Management on New Hardware - Seattle, Washington, USA, Seattle, United States Duration: 19 Jun 2023 → 19 Jun 2023 https://www.damon-db.org/

Conference

Conference	Conference: International Workshop on Data Management on New Hardware
Location	Seattle, Washington, USA
Country/Territory	United States
City	Seattle
Period	19/06/2023 → 19/06/2023
Internet address	https://www.damon-db.org/

Keywords

Computational storage
NVMe namespaces
eBPF
Code offloading
Storage device performance

Access to Document

10.1145/3592980.3595319Licence: CC BY

damon23-20Final published version, 1.47 MBLicence: CC BY

Cite this

@inproceedings{d6484671b01d4954a04ef594e27403fe,

title = "Delilah: eBPF-offload on Computational Storage",

abstract = "The idea of pushing computation to storage devices has been explored for decades, without widespread adoption so far. The definition of Computational Programs namespaces in NVMe (TP 4091) might be a breakthrough. The proposal defines device-specific programs, that are installed statically, and downloadable programs, offloaded from a host at run-time using eBPF. In this paper, we present the design and implementation of Delilah, the first public description of an actual computational storage device supporting eBPF-based code offload. We conduct experiments to evaluate the overhead of eBPF function execution in Delilah, and to explore design options. This study constitutes a baseline for future work.",

keywords = "Computational storage, NVMe namespaces, eBPF, Code offloading, Storage device performance, Computational storage, NVMe namespaces, eBPF, Code offloading, Storage device performance",

author = "Niclas Hedam and Clausen, {Morten Tychsen} and Philippe Bonnet and Sangjin Lee and Larsen, {Ken Friis}",

year = "2023",

month = jun,

day = "19",

doi = "10.1145/3592980.3595319",

language = "English",

booktitle = "International Workshop on Data Management on New Hardware (DaMoN)",

note = "Conference: International Workshop on Data Management on New Hardware ; Conference date: 19-06-2023 Through 19-06-2023",

url = "https://www.damon-db.org/",

}

TY - GEN

T1 - Delilah: eBPF-offload on Computational Storage

AU - Hedam, Niclas

AU - Clausen, Morten Tychsen

AU - Bonnet, Philippe

AU - Lee, Sangjin

AU - Larsen, Ken Friis

PY - 2023/6/19

Y1 - 2023/6/19

N2 - The idea of pushing computation to storage devices has been explored for decades, without widespread adoption so far. The definition of Computational Programs namespaces in NVMe (TP 4091) might be a breakthrough. The proposal defines device-specific programs, that are installed statically, and downloadable programs, offloaded from a host at run-time using eBPF. In this paper, we present the design and implementation of Delilah, the first public description of an actual computational storage device supporting eBPF-based code offload. We conduct experiments to evaluate the overhead of eBPF function execution in Delilah, and to explore design options. This study constitutes a baseline for future work.

AB - The idea of pushing computation to storage devices has been explored for decades, without widespread adoption so far. The definition of Computational Programs namespaces in NVMe (TP 4091) might be a breakthrough. The proposal defines device-specific programs, that are installed statically, and downloadable programs, offloaded from a host at run-time using eBPF. In this paper, we present the design and implementation of Delilah, the first public description of an actual computational storage device supporting eBPF-based code offload. We conduct experiments to evaluate the overhead of eBPF function execution in Delilah, and to explore design options. This study constitutes a baseline for future work.

KW - Computational storage

KW - NVMe namespaces

KW - eBPF

KW - Code offloading

KW - Storage device performance

KW - Computational storage

KW - NVMe namespaces

KW - eBPF

KW - Code offloading

KW - Storage device performance

UR - https://www.researchgate.net/profile/Niclas-Hedam/publication/370818368_Delilah_eBPF-offload_on_Computational_Storage/links/6464a6b29533894cac7662d3/Delilah-eBPF-offload-on-Computational-Storage.pdf

U2 - 10.1145/3592980.3595319

DO - 10.1145/3592980.3595319

M3 - Article in proceedings

BT - International Workshop on Data Management on New Hardware (DaMoN)

CY - Seattle, Washington, USA

T2 - Conference: International Workshop on Data Management on New Hardware

Y2 - 19 June 2023 through 19 June 2023

ER -

Delilah: eBPF-offload on Computational Storage

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this