RAFlow: Read ahead accelerated I/O flow through multiple virtual layers

Zhaoning Zhang; Kui Wu; Huiba Li; Jinghua Feng; Yuxing Peng; Xicheng Lu

doi:10.1109/NAS.2014.13

RAFlow: Read ahead accelerated I/O flow through multiple virtual layers

Zhaoning Zhang^*, Kui Wu, Huiba Li, Jinghua Feng, Yuxing Peng, Xicheng Lu

^*Corresponding author for this work

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

1 Citation (Scopus)

Abstract

Virtualization is the foundation for cloud computing, and the virtualization can not be achieved without software defined, elastic, flexible and scalable virtual layers. Unfortunately, if multiple virtual storage devices are chained together, the system may be subject to severe performance degradation. While the read-ahead (RA) mechanism in storage devices plays a very important role to improve I/O performance, RA may not be effective as expected for multiple virtualization layers, since it is originally designed for one layer only. When I/O requests are passed through a long I/O path, they may trigger a chain reaction and lead to unnecessary data transmission and thus bandwidth waste. In this paper, we study the dynamic behavior of RA through multiple I/O layers and demonstrate that if controlled well, RA can greatly accelerate I/O speed. We present RAFlow, a RA control mechanism, to effectively improve I/O performance by strategically expanding RA window at each layer. Our real-world experiments show that it can achieve 20% to 50% performance improvement in I/O paths with up to 8 virtualized storage devices. © 2014 IEEE.

Original language	English
Title of host publication	Proceedings - 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014
Publisher	IEEE
Pages	33-42
ISBN (Print)	9781479940875
DOIs	https://doi.org/10.1109/NAS.2014.13
Publication status	Published - 13 Oct 2014
Externally published	Yes
Event	9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014 - Tianjin, China Duration: 6 Aug 2014 → 8 Aug 2014

Conference

Conference	9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014
Place	China
City	Tianjin
Period	6/08/14 → 8/08/14

Bibliographical note

Publication details (e.g. title, author(s), publication statuses and dates) are captured on an “AS IS” and “AS AVAILABLE” basis at the time of record harvesting from the data source. Suggestions for further amendments or supplementary information can be sent to [email protected].

Research Keywords

Cloud Computing
I/O Performance
IaaS Storage
Read ahead

Access Output

10.1109/NAS.2014.13

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@inproceedings{74acc0ea15ad4a75be787f3dc140613b,

title = "RAFlow: Read ahead accelerated I/O flow through multiple virtual layers",

abstract = "Virtualization is the foundation for cloud computing, and the virtualization can not be achieved without software defined, elastic, flexible and scalable virtual layers. Unfortunately, if multiple virtual storage devices are chained together, the system may be subject to severe performance degradation. While the read-ahead (RA) mechanism in storage devices plays a very important role to improve I/O performance, RA may not be effective as expected for multiple virtualization layers, since it is originally designed for one layer only. When I/O requests are passed through a long I/O path, they may trigger a chain reaction and lead to unnecessary data transmission and thus bandwidth waste. In this paper, we study the dynamic behavior of RA through multiple I/O layers and demonstrate that if controlled well, RA can greatly accelerate I/O speed. We present RAFlow, a RA control mechanism, to effectively improve I/O performance by strategically expanding RA window at each layer. Our real-world experiments show that it can achieve 20% to 50% performance improvement in I/O paths with up to 8 virtualized storage devices. {\textcopyright} 2014 IEEE.",

keywords = "Cloud Computing, I/O Performance, IaaS Storage, Read ahead",

author = "Zhaoning Zhang and Kui Wu and Huiba Li and Jinghua Feng and Yuxing Peng and Xicheng Lu",

note = "Publication details (e.g. title, author(s), publication statuses and dates) are captured on an “AS IS” and “AS AVAILABLE” basis at the time of record harvesting from the data source. Suggestions for further amendments or supplementary information can be sent to [email protected].; 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014 ; Conference date: 06-08-2014 Through 08-08-2014",

year = "2014",

month = oct,

day = "13",

doi = "10.1109/NAS.2014.13",

language = "English",

isbn = "9781479940875",

pages = "33--42",

booktitle = "Proceedings - 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014",

publisher = "IEEE",

address = "United States",

}

Zhang, Z, Wu, K, Li, H, Feng, J, Peng, Y & Lu, X 2014, RAFlow: Read ahead accelerated I/O flow through multiple virtual layers. in Proceedings - 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014., 6923156, IEEE, pp. 33-42, 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014, Tianjin, China, 6/08/14. https://doi.org/10.1109/NAS.2014.13

RAFlow: Read ahead accelerated I/O flow through multiple virtual layers. / Zhang, Zhaoning; Wu, Kui; Li, Huiba et al.
Proceedings - 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014. IEEE, 2014. p. 33-42 6923156.

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - RAFlow

T2 - 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014

AU - Zhang, Zhaoning

AU - Wu, Kui

AU - Li, Huiba

AU - Feng, Jinghua

AU - Peng, Yuxing

AU - Lu, Xicheng

N1 - Publication details (e.g. title, author(s), publication statuses and dates) are captured on an “AS IS” and “AS AVAILABLE” basis at the time of record harvesting from the data source. Suggestions for further amendments or supplementary information can be sent to [email protected].

PY - 2014/10/13

Y1 - 2014/10/13

N2 - Virtualization is the foundation for cloud computing, and the virtualization can not be achieved without software defined, elastic, flexible and scalable virtual layers. Unfortunately, if multiple virtual storage devices are chained together, the system may be subject to severe performance degradation. While the read-ahead (RA) mechanism in storage devices plays a very important role to improve I/O performance, RA may not be effective as expected for multiple virtualization layers, since it is originally designed for one layer only. When I/O requests are passed through a long I/O path, they may trigger a chain reaction and lead to unnecessary data transmission and thus bandwidth waste. In this paper, we study the dynamic behavior of RA through multiple I/O layers and demonstrate that if controlled well, RA can greatly accelerate I/O speed. We present RAFlow, a RA control mechanism, to effectively improve I/O performance by strategically expanding RA window at each layer. Our real-world experiments show that it can achieve 20% to 50% performance improvement in I/O paths with up to 8 virtualized storage devices. © 2014 IEEE.

AB - Virtualization is the foundation for cloud computing, and the virtualization can not be achieved without software defined, elastic, flexible and scalable virtual layers. Unfortunately, if multiple virtual storage devices are chained together, the system may be subject to severe performance degradation. While the read-ahead (RA) mechanism in storage devices plays a very important role to improve I/O performance, RA may not be effective as expected for multiple virtualization layers, since it is originally designed for one layer only. When I/O requests are passed through a long I/O path, they may trigger a chain reaction and lead to unnecessary data transmission and thus bandwidth waste. In this paper, we study the dynamic behavior of RA through multiple I/O layers and demonstrate that if controlled well, RA can greatly accelerate I/O speed. We present RAFlow, a RA control mechanism, to effectively improve I/O performance by strategically expanding RA window at each layer. Our real-world experiments show that it can achieve 20% to 50% performance improvement in I/O paths with up to 8 virtualized storage devices. © 2014 IEEE.

KW - Cloud Computing

KW - I/O Performance

KW - IaaS Storage

KW - Read ahead

UR - http://www.scopus.com/inward/record.url?scp=84912027591&partnerID=8YFLogxK

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-84912027591&origin=recordpage

U2 - 10.1109/NAS.2014.13

DO - 10.1109/NAS.2014.13

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 9781479940875

SP - 33

EP - 42

BT - Proceedings - 9th IEEE International Conference on Networking, Architecture, and Storage, NAS 2014

PB - IEEE

Y2 - 6 August 2014 through 8 August 2014

ER -

RAFlow: Read ahead accelerated I/O flow through multiple virtual layers

Abstract

Conference

Bibliographical note

Research Keywords

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this