Lead or attend a BoF! Meet with your peers! Present new work! Don't miss these special activities designed to maximize the value of your time at the conference. The always popular evening Birds-of-a-Feather sessions are very informal gatherings of persons interested in a particular topic.
Vendor BoFs
Want to demonstrate a new product or discuss your company's latest technologies with FAST '16 attendees? Host a Vendor BoF! These sponsored one-hour sessions give companies a chance to talk about products and proprietary technology—and they include promotional benefits. Email sponsorship@usenix.org if you're interested in sponsoring a Vendor BoF. Click here for more information about sponsorship opportunities.
Scheduling a BoF
It's not too late! To schedule a BoF, simply write the BoF title as well as your name and affiliation on one of the BoF Boards located in the registration area. If you have a description of your BoF you'd like posted on this Web page, please schedule your BoF on the BoF board, then send its title, the organizer's name and affiliation, and the date, time, and location of the BoF to bofs@usenix.org with "FAST 16 BoF" in the subject line.
SNIA NVM Programming Model
Doug Voigt, NVM Programming Model TWG Chair, Hewlett Packard Enterprise
Tuesday, February 23, 8:00 pm–9:00 pm, Alameda Room
The SNIA NVM Programming model describes behaviors that application, OS and hardware vendors can use to form a diverse yet interoperable NVM ecosystem. This is especially important as post-flash technologies enable new application behaviors with Storage Class Memory. We will briefly review the model and discuss recent related developments in the areas of atomicity, recoverability and remote access for high availability.
File System and Storage Benchmarking and Tracing: On Formalizing Filebench Workloads
Erez Zadok, Stony Brook University
Vasily Tarasov, IBM Research
Geoff Kuenning, Harvey Mudd College and SNIA
Tuesday, February 23, 8:00 pm–9:00 pm, Grand Ballroom DE
File system and storage benchmarking is an area with many open and flammable debates. Often, the evaluation section of a paper is the most heavily criticized section of a research submitted to FAST and other storage venues. Were the workloads picked appropriately? Are they sufficiently representative? Was the right dataset size used? Is the methodology used statistically and scientifically sound? Different people often have very different answers to these questions. The lack of a baseline impedes a fair and unbiased comparison between different solutions. In this BoF we will first summarize the status of the SNIA IOTTA Trace Repository, then turn to Filebench and discuss some future plans. We note that Filebench's benchmarking framework is quite popular in the community and might provide a useful baseline for accurately describing various micro- and macro-workloads. We then invite everyone to an open discussion on how the quality of benchmarking can be improved. Ideally, we would like to establish specific steps that the community can take to promote better benchmarking practices. Specifically, we are looking for establishing an expert community for defining and maintaining modern, relevant, and diverse workloads in Filebench on an ongoing basis.
Tintri Vendor BoF: Designing for Flash-Based Storage: An Industry Perspective Brandon Salmon, Tintri
Tuesday, February 23, 9:00 pm–10:00 pm, Grand Ballroom ABCFGH
Flash storage has radically different characteristics than disk. For this reason, flash-based storage systems have radically different architectures than traditional disk-based storage. Because there are currently few open-source flash-based storage systems, it is often difficult for researchers to develop advances that fit well within a flash storage architecture.
In this BoF we will discuss some of the key architectural differences between flash-based and disk-based storage systems, drawn from experience building a widely deployed Enterprise flash-based storage system at Tintri. We will also discuss the impact this has on storage research, and how to design advances that will fit well in modern flash-based storage. Light refreshments will be served.
CloudPhysics 2015 Traces & Miss Ratio Curves For Storage Irfan Ahmad & Carl Waldspurger, CloudPhysics
Wednesday, February 24, 8:00 pm–10:00 pm, Grand Ballroom ABCFGH
(1) CloudPhysics is in the process of releasing traces used in the SHARDS (FAST '15) paper. These are over 100 traces from real production, customer environments from dozens of different data centers of different companies. All traces are at least a week long. In this BoF, we'll discuss the traces, logistics for releasing the traces, appropriate license, etc. Please attend if you'd like to influence any of these decisions.
(2) MRC computation has traditionally been too resource-intensive to be broadly practical. Recent works (FAST '15, OSDC '14) have made online MRCs practical. At this BoF session, we will have an open discussion on the directions online cache management can take, with a particular emphasis on translating research into practical systems.
Red Hat Vendor BoF: Gluster.Next Vijay Bellur, Project Lead, Gluster
Wednesday, February 24, 8:00 pm–9:00 pm, Grand Ballroom DE
Gluster.Next refers to the ongoing architectural evolution in Gluster to scale better and enable new use cases. In this discussion, Vijay will talk about slew of improvements planned for Gluster.next - DHTv2, NSR, Glusterd 2.0, Heketi, Brick multiplexing, Quality of Service amongst others.
Use cases like storage as a service, storage for containers and hyperconvergence being enabled by this new wave of Gluster will be discussed. Integrations with other projects like OpenShift, OpenStack and oVirt will also be highlighted in this presentation. Light refreshments will be served.
Twitter
Facebook
LinkedIn
Google+
YouTube
connect with us