Skip to main content
USENIX
  • Conferences
  • Students
Sign in
  • Home
  • Attend
    • Registration Information
    • Registration Discounts
    • Venue, Hotel, and Travel
    • Students and Grants
  • Program
    • At a Glance
    • Technical Sessions
    • Training Program
    • Poster Sessions
    • WiPs
  • Activities
    • Birds-of-a-Feather Sessions
    • Poster Sessions
  • Sponsorship
  • Participate
    • Call for Papers
    • Call for Posters and WiPs
    • Instructions for Participants
  • About
    • Conference Organizers
    • Questions
    • Services
    • Help Promote!
    • Past Conferences
  • Home
  • Attend
  • Program
  • Activities
  • Sponsorship
  • Participate
  • About

sponsors

Platinum Sponsor
Gold Sponsor
Gold Sponsor
Gold Sponsor
Gold Sponsor
Silver Sponsor
Silver Sponsor
Silver Sponsor
Silver Sponsor
Silver Sponsor
Bronze Sponsor
Bronze Sponsor
Bronze Sponsor
Bronze Sponsor
Bronze Sponsor
Media Sponsor
Media Sponsor
Media Sponsor
Media Sponsor
Media Sponsor
Media Sponsor
Media Sponsor
Media Sponsor
Industry Partner
Industry Partner

help promote

FAST '17 CFP

Get
Help Promote graphics!

connect with us


  •  Twitter
  •  Facebook
  •  LinkedIn
  •  Google+
  •  YouTube

twitter

Tweets by @usenix

usenix conference policies

  • Event Code of Conduct
  • Conference Network Policy
  • Statement on Environmental Responsibility Policy

You are here

Home » Using Hints to Improve Inline Block-layer Deduplication
Tweet

connect with us

Using Hints to Improve Inline Block-layer Deduplication

Authors: 

Sonam Mandal, Stony Brook University; Geoff Kuenning, Harvey Mudd College; Dongju Ok and Varun Shastry, Stony Brook University; Philip Shilane, EMC Corporation; Sun Zhen, Stony Brook University and National University of Defense Technology; Vasily Tarasov, IBM Research; Erez Zadok, Stony Brook University

Abstract: 

Block-layer data deduplication allows file systems and applications to reap the benefits of deduplication without requiring per-system or per-application modifications. However, important information about data context (e.g., data vs. metadata writes) is lost at the block layer. Passing such context to the block layer can help improve deduplication performance and reliability. We implemented a hinting interface in an open-source block-layer deduplication system, dmdedup, that passes relevant context to the block layer, and evaluated two hints, NODEDUP and PREFETCH. To allow upper storage layers to pass hints based on the available context, we modified the VFS and file system layers to expose a hinting interface to user applications. We show that passing the NODEDUP hint speeds up applications by up to 5.3 on modern machines because the overhead of deduplication is avoided when it is unlikely to be beneficial. We also show that the PREFETCH hint accelerates applications up to 1.8 by caching hashes for data that is likely to be accessed soon.

Sonam Mandal, Stony Brook University

Geoff Kuenning, Harvey Mudd College

Dongju Ok, Stony Brook University

Varun Shastry, Stony Brook University

Philip Shilane, EMC Corporation

Sun Zhen, Stony Brook University and National University of Defense Technology

Vasily Tarasov, IBM Research

Erez Zadok, Stony Brook University

Open Access Media

USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.

BibTeX
@inproceedings {194452,
author = {Sonam Mandal and Geoff Kuenning and Dongju Ok and Varun Shastry and Philip Shilane and Sun Zhen and Vasily Tarasov and Erez Zadok},
title = {Using Hints to Improve Inline Block-layer Deduplication},
booktitle = {14th USENIX Conference on File and Storage Technologies (FAST 16)},
year = {2016},
isbn = {978-1-931971-28-7},
address = {Santa Clara, CA},
pages = {315--322},
url = {https://www.usenix.org/conference/fast16/technical-sessions/presentation/mandal},
publisher = {USENIX Association},
month = feb,
}
Download
Mandal PDF
View the slides

Presentation Audio

MP3 Download

Download Audio

  • Log in or    Register to post comments

Platinum Sponsors

Gold Sponsors

Silver Sponsors

Bronze Sponsors

Media Sponsors & Industry Partners

Open Access Publishing Partner

© USENIX

  • Privacy Policy
  • Contact Us