Hosted at

32nd International Conference
on Massive Storage Systems
and Technology (MSST 2016)
May 2 — 6, 2016

Sponsored by Santa Clara University,
School of Engineering

Since the conference was founded by the leading national laboratories, MSST has been a venue for massive-scale storage system designers and implementers, storage architects, researchers, and vendors to share best practices and discuss building and securing the world's largest storage systems for high-performance computing, web-scale systems, and enterprises.

Technically Co-
Sponsored by

Hosted at
Santa Clara University
Santa Clara, CA

Invited Track
Research Track

2016 Conference

MSST (2016), as is our custom, dedicated five days to computer-storage technology, including a day of tutorials, two days of invited papers, two days of peer-reviewed research papers, and a vendor exposition. The conference was held, once again, on the beautiful campus of Santa Clara University, in the heart of Silicon Valley.

InsideHPC Video Gallery of Invited Talks

Santa Clara University

Why Workloads Matter More Than
IOPS or Streaming
, by Henry Newman

Data-Intensive Workflows: Take 2
by Matthew O’Keefe

Interview: Jeff Bonwick on the
Secret Sauce behind DSSD

Interview: Gary Grider on Moving
Beyond POSIX with the new
MarFS Object Storage Project

Many Thanks to Our Sponsors!

SGI logo            Oracle logo
Cray logo Seagate logo

SpectraLogic logo

2016 Program

Tutorial, Monday, May 2nd
7:30 — 8:30 Registration / Breakfast
8:30 — 5:00 Swiftstack Swift Tutorial / Lunch
Albert Chen, Systems Engineer, Swiftstack
Chris Nelson, Director of Systems Engineering, Swiftstack
Abstract: In this full-day workshop, you'll learn why the largest public clouds and internet properties, including Wikipedia, IBM Softlayer, Rackspace, and HP are using OpenStack Swift to deliver massively scalable, multi-geographically distributed storage for their customers. You'll also learn how to get up and running with a Swift Cluster.

In this hands-on lab, we'll cover: More specifically, you'll leave understanding exactly how to deploy and manage a Swiftcluster, what use cases are the best fit for OpenStack Swift's object storage and how easy it is to build apps using Swift and consume their assets (videos, images, docs, pdfs, etc...).

Prerequisites: Attendees will be given access to their own remote linux instances for the hands-on lab. Please make sure you have the following for the tutorial session:

Invited Track, Tuesday, May 3rd
7:30 — 8:30 Registration / Breakfast
8:30 — Keynote
Scalable High Performance FLASH Systems (Slides, Video)
Jeff Bonwick, EMC DSSD
Jeff Bonwick
MSST 2016 features keynote speaker Jeff Bonwick, co-founder and CTO of DSSD, where he co-invented both the system hardware architecture and the Flood software stack. His talk will focus on extracting maximum performance from flash at scale. Jeff has a long history of developing at-scale storage starting with leading the team that developed the ZFS filesystem, which powers Oracle’s ZFS storage line as well as numerous startups including Nexenta, Delphix, Joyent, and Datto. Jeff is also the inventor of the slab allocator, which is used in every major operating system and has become part of the computer science curriculum at universities worldwide. His previous roles include Sun Fellow, Sun storage CTO, and Oracle VP. Jeff holds 62 issued patents in storage and operating system technology.
Application- and Workload-Specific Workflows
Workflow Specification for Large-Scale Computational Systems (Slides, Video)
David Montoya, Los Alamos National Laboratory
Panel: Workflows for Specifying System Acquisitions
Moderator: Gary Grider, Los Alamos National Laboratory (Slides, Video)
David Montoya, Los Alamos National Laboratory
Yoonho Park, IBM (Slides, Video)
Bret Weber, DDN (Slides, Video)
Lance Evans, Cray, Inc. (Slides)
12:30 — 1:30 Lunch
1:30 — Data-Intensive Workflows
A Holistic Framework for Data-Intensive Workflows
Ian Corner, CSIRO (Australia) (Slides, Video)
Panel: Data-Intensive Workflows
Moderator: Ian Corner, CSIRO (Australia) (Panel Introduction, Panel Video)
Jake Carroll, University of Queensland Brain Institute (Slides)
Kirk Jordan, Hartree Centre - Science and Technology Facilities Council
Massimo Noro, Unilever (Slides)
Don Preuss, Starfish Storage
Leveraging Disk for Large-Scale, Long-Term Storage Applications
A Perspective on Power Management for Hard Disks
Kirill Malkin, SGI (Slides)
Kirill Malkin

The speaker will talk about the impact of power management on reliability of hard disk devices based on a field experience with SGI’s MAID product (formerly COPAN). The presentation will explore statistical analysis of support data covering a decade of MAID deployment (2005-2015) as well as briefly cover the history and basic principles of the technology and provide an update on future direction of the product line.

Hard Disks for Large Archives
Dave Anderson, Seagate
Leveraging Large-Scale Disk Systems in a Web-based Photo Archive: Design, Operations, and Future Plans
Mike Kugler, Shutterfly (Slides)
Panel: Leveraging Disks for Large-Scale, Long-Term Storage
Moderator: Jim Gerry, IBM (Panel Introduction)
Dave Anderson, Seagate (Slides, Video)
Roark Hilomen, Sandisk (Video)
Mike Kugler, Shutterfly (Video)
Kirill Malkin, SGI (Video)
Nilay Patel, Backblaze (Slides, Video)
Lightning Talks
Superfacility: How new workflows in the DOE Office of Science are changing storage system requirements
Katie Antypas, Lawrence Berkeley Laboratory (Slides, Video)
Everspan—an optical archive solution with LTO-7 performance and cost, but without the tape issues
Horst Schellong, Sony (Slides, Video)
Invited Track, Wednesday, May 4th
7:30 — 8:30 Breakfast
8:30 — Keynote
Learnings from Operating 200 PB of Disk-Based Storage
Gleb Budman, Backblaze (Slides, Video)
Trends in FLASH Technology Relevant to Large-Scale Systems
Leveraging Flash in Scalable Environments: A Systems Perspective on How FLASH Storage is Displacing Disk Storage
Roark Hilomen, Sandisk (Slides, Video)
Storage Media Overview: Historic Perspectives
Bob Fontana, IBM Almaden Research (Slides, Video)
Innovations in Non-Volatile Memory: 3D NAND and its Implications
Rob Peglar, Micron (Slides, Video)
Panel: Media Trends
Moderator: Matthew O’Keefe, Oracle
April Alstrin, Oracle (Slides, Video)
Dave Anderson, Seagate (Slides)
Bob Fontana, IBM Almaden Research
Roark Hilomen, Sandisk
Rob Peglar, Micron
12:30 — 1:30 Lunch
1:30 — An Adventure in Semantics: Compromising POSIX Semantics to Adapt
Applications to Scalable Object Technology
MarsFS: A Near-POSIX Namespace Leveraging Scalable Object Storage (Slides, Video)
David Bonnie, Los Alamos National Laboratory
David Bonnie
It has been said that objects are for applications and POSIX is for people. The HPC community as well as many other large scale IT organizations have legacy applications and users that know, use, and depend on a near POSIX environment with real folders, ease of renaming and reshaping trees, and other powerful concepts in POSIX.

There are several POSIX name spaces that sit on top of cloud style erasure based objects but few if any really provide an extremely scalable solution. MarFS is designed to address this problem by providing a scalable near-POSIX name space over standard object systems, with target scaling out to trillions of POSIX files, hundreds of gigabytes/sec of data bandwidth, and millions of POSIX metadata operations/sec.
Panel: Evolving Semantics for Object Storage
Moderator: Randy Olinger, Optum United Health Group (Slides, Video, Discussion Video)
David Bonnie, Los Alamos National Laboratory (Slides)
Ian Corner, CSIRO
Harriet Coverston, Versity Software (Slides, Video)
Carrie Spear, NASA Goddard (Slides, Video)
Justin Stottlemyer, Intuit
Eternal 5D Data Storage in Glass
Peter Kazansky, University of Southampton (UK) (Slides, Video)
Recording of polarization multiplexed digital data was demonstrated by femtosecond laser nanostructuring of fused quartz. The storage allows unprecedented parameters including hundreds of terabytes per disc data capacity, thermal stability up to 1000°C and virtually unlimited lifetime at room temperature, which is a vital step towards an eternal archive.
Everspan: 8-channel optical drive to achieve high transfer rates
Horst Schellong, Sony (Slides, Video)
Building Scalable, High Performance Block Storage via an RDMA-based Hyperconverged Platform
Josh Goldenhar, Excelero
Excelero NVMesh allows the seemingly impossible by enabling compute hosts to have shared access to 100s of TBs of flash storage at LOCAL speeds and latencies. By incorporating classic RDMA benefits with NVMe technology and a few new “tricks”, it’s become possible to build a hugely scalable distributed block platform that performs faster than any all-flash-array. This capability allows data scientists to slice and dice data for analysis in ways that before were too costly, too slow or too proprietary.
5:30 — 8:00 Reception

Research Track, Thursday, May 5th
(* Indicates Presenter)
8:00 — 9:00 Registration / Breakfast
9:00 — It’s Never too Fast: Storage Performance Enhancements
Session chair: Dimitris Skourtis, VMware
Pfimbi: Accelerating Big Data Jobs Through Flow-Controlled Data Replication (Paper, Slides)
Simbarashe Dzinamarira* and T. S. Eugene Ng, Rice University
Florin Dinu, École Polytechnique Fédérale de Lausanne
Manylogs: Improved CMR/SMR Disk Bandwidth and Faster Durability with Scattered Logs (Paper, Slides)
Tiratat Patana-anake*, Nora Sandler, Cheng Wu and Haryadi S. Gunawi, University of Chicago
Vincentius Martin, Surya University
Tombolo: Performance Enhancements for Cloud Storage Gateways (Paper, Slides)
Suli Yang*, Remzi H. Arpaci-Dusseau, Andrea C. Arpaci-Dusseau, University of Wisconsin
Kiran Srinivasan, Kishore Udayashankar, Shweta Krishnan, Jingxin Feng, NetApp
Yupu Zhang, HP
File Systems for Non-Volatile Memory
Session chair: Jay Lofstead, Sandia National Laboratories
Fine-grained Metadata Journaling on NVM (Paper, Slides)
Cheng Chen, Jun Yang*, Qingsong Wei, Chundong Wang, and Mingdi Xue, A-STAR
Fast and Failure-Consistent Updates of Application Data in Non-Volatile Main Memory File System (Paper, Slides)
Jiaxin Ou and Jiwu Shu, Tsinghua University, presented by Linpeng Huang*
HMVFS: A Hybrid Memory Versioning File System (Paper, Slides)
Shengan Zheng*, Linpeng Huang, Hao Liu, Linzhu Wu, and Jin Zha, Shanghai Jiao Tong University
12:10 — 1:10 Lunch
1:10 — Store More, Longer, and for Less: Deduplication and Archival Systems
Session chair: John May, Lawrence Livermore National Laboratory
A Long Term User-Centric Analysis of Deduplication Patterns (Paper, Slides)
Zhen Sun* and Nong Xiao, National University of Defense Technology
Geoff Kuenning, Harvey Mudd College
Sonam Mandal and Erez Zadok, Stony Brook University
Philip Shilane, EMC
Vasily Tarasov, IBM Research
Lazy Exact Deduplication (Paper, Slides)
Jingwei Ma, Rebecca J. Stones*, Yuxiang Ma, Jingui Wang, Junjie Ren, Gang Wang, Xiaoguang Liu, Nankai University
Sorted Deduplication: How to Process Thousands of Backup Streams (Paper, Slides)
Jürgen Kaiser*, Tim Süß, Lars Nagel, André Brinkmann, Johannes Gutenberg University
Effects of Prolonged Media Usage and Long-term Planning on Archival Systems (Paper, Slides)
Preeti Gupta, Darrell D.E. Long, and Ethan L. Miller, University of California, Santa Cruz
Avani Wildani*, Emory University
David S.H. Rosenthal, Stanford University
Spotlight on Flash memory and Solid-State Drives
Session chair: Irfan Ahmad, Cloud Physics
Adaptive policies for balancing performance and lifetime of mixed SSD arrays through workload sampling
(Paper, Slides)
Sangwhan Moon* and A. L. Narasimha Reddy, Texas A & M University
REAL: A Retention Error Aware LDPC Decoding Scheme to Improve NAND Flash Read Performance (Paper, Slides)
Meng Zhang*, Fei Wu, Shunzhuo Wang, and Changsheng Xie, Huazhong University of Science and Technology
Xubin He and Ping Huang, Virginia Commonwealth University
Analytic models for flash-based SSD performance when subject to trimming (Paper, Slides)
Robin Verschoren* and Benny Van Houdt, University of Antwerp
Reducing Write Amplification of Flash Storage through Cooperative Data Management with NVM (Paper, Slides)
Eunji Lee*, Chungbuk National University
Julie Kim, Line Corporation
Hyokyung Bahn, Ewha Womans University
Sam H. Noh, Ulsan National Institute of Science and Technology
Exploiting Latency Variation for Access Conflict Reduction of NAND Flash Memory (Paper, Slides)
Jinhua Cui, Weiguo Wu, Xingjun Zhang, and Jianhang Huang*, Xi'an Jiaotong University
Yinfeng Wang, ShenZhen Institute of Information Technology
Research Track, Friday, May 6th
(* Indicates Presenter)
8:00 — 9:00 Breakfast
9:00 — Understanding Storage Systems through Measurements and Analysis
Session chair: Xing Lin, NetApp
Understanding I/O Performance Behaviors of Cloud Storage from a Client’s Perspective (Paper, Slides)
Binbing Hou* and Feng Chen, Louisiana State University
Zhonghong Ou, Beijing University of Posts and Telecommunications
Ren Wang and Michael Mesnier, Intel Labs
File System Trace Replay Methods Through the Lens of Metrology (Paper, Slides)
Thiago Emmanuel Pereira*, Francisco Brasileiro, and Livia Sampaio , Universidade Federal de Campina Grande
The Impact of Data Placement on Resilience in Large-Scale Object Storage Systems (Paper, Slides)
Philip Carns*, Kevin Harms, John Jenkins, Misbah Mubarak, Robert Ross, Argonne National Laboratory
Christopher Carothers, Rensselaer Polytechnic Institute
On-the-Go Storage
Session chair: Jishen Zhao, University of California, Santa Cruz
Understanding Storage I/O Behaviors of Mobile Applications (Paper, Slides)
Jace Courville* and Feng Chen, Louisiana State University
An Overlay File System for Cloud-Assisted Mobile Applications (Paper, Slides)
Jianchen Shan*, Nafize R. Paiker, Xiaoning Ding, Narain Gehani, Reza Curtmola and Cristian Borcea, New Jersey Institute of Technology
Fast Transaction Logging for Smartphones (Paper, Slides)
Hao Luo*, Yaodong Yang, University of Nebraska
Hong Jiang, Zhichao Yan, University of Texas
12:00 — 1:00 Lunch

2016 Organizers
Advisory Board
Conference Chair     Dr. Sam Coleman
Tutorial Chair     Sean Roberts
Program Chair     Dr. Matthew O'Keefe
Research General Chair     Dr. Ahmed Amer
                Research Program Chairs     Dr. Carlos Maltzahn, Dr. Vasily Tarasov
Research Track Program Committee
SCU Arrangements     Dr. Ahmed Amer
Vendor Chair     Dr. James Reaney
Communications Chair     Meghan Wingate McClelland
Registration Chairs     Dr. JoAnne Holliday, Yi Fang

Page Updated January 12, 2024