Vault 2016 has ended
Dev Focus [clear filter]
Wednesday, April 20

10:30am PDT

Open-Channel Solid State Drives - Matias Bjørling CNEX Labs
This presentation gives an introduction to Open-Channels SSDs and LightNVM - the kernel subsystem for Open-Channel SSDs. Open-Channel SSDs allow applications to directly control data placement and garbage collection within an SSD. The benefits are consistent latency and throughput in high-performance workloads and extending SSD lifetime. The primary objective of this talk is to show the possibilities of this new storage interface, its architecture, the current status and roadmap for LightNVM for both upcoming kernel functionality and user-space support.


Matias Bjørling

Director, Western Digital
Matias Bjørling researchs emerging storage architectures Western Digital. Before joining the industry, he obtained a Ph.D. in operating systems, and non-volatile storage by doing performance characterization of flash-based SSDs, working on thr linux kernel blk-mq, and its associated... Read More →

Wednesday April 20, 2016 10:30am - 11:20am PDT
State Ballroom A

11:30am PDT

Managing Fabric-Attached Memory in The Machine - Rocky Craig, Hewlett Packard Enterprise
Fabric-Attached Memory (Rocky Craig, Hewlett Packard Enterprise) - The Machine from HPE is the prototoype of a novel architecture featuring memory-centric computing. Multiple Linux nodes see fabric-attached NVM in their physical address space as a shared global resource. This presentation introduces the hardware architecture as it relates to the NVM and the challenges in shared management. While NVM presents a key-value paradigm in data retrieval, we chose the route of a global file system. The bulk of the talk will focus on this "Librarian File System" and the benefits of a well-known API versus a proprietary solution.

avatar for Rocky Craig

Rocky Craig

Principal Lead, Hewlett Packard Enterprise
Rocky is employed by Hewlett Packard Enterprise, a spinoff of HP where he started in 1978. He has served many roles in the technical space of technical workstations and HP-UX and Linux servers. Rocky is currently co-architect of a Debian-based distribution for The Machine, a new architecture... Read More →

Wednesday April 20, 2016 11:30am - 12:20pm PDT
State Ballroom A

2:00pm PDT

SMB3 in Samba - Multi-Channel and Beyond - Michael Adam, Red Hat
Samba is the most important open source SMB file server, and arguably, one of the most important SMB
implementations out there. While the Active Directory Server has been the hip topic recently, the file server is really cooking. One of the most active areas of development is SMB version 3, introduced by Microsoft to move focus from pure workstation workload to server workload. SMB3 adds an abundance of new features to the protocol. In particular, Microsoft caught up with Samba/CTDB to offer all-active clustering.

After an overview of the state of Samba's file server, this talk will present the progress of SMB3 in Samba. It will focus on Multi-Channel, the core of which has just been added to Samba 4.4, including a demo. Afterwards an outlook will be given on other active areas like RDMA, persistent file handles, and scale-out SMB clustering, reporting on status and challenges.

avatar for Michael Adam

Michael Adam

Principal Software Engineer, Red Hat
Michael Adam is an enthusiastic open source software developer, interested in all things about storage and containers. One of the main developers of Samba since more than a decade, Michael is an engineering manager at Red Hat, leading two worldwide teams: The Samba team for Gluster... Read More →

Wednesday April 20, 2016 2:00pm - 2:50pm PDT
State Ballroom A

3:00pm PDT

MD/RAID-456 Write Journal and Cache - Song Liu, Facebook
Without NVRAM and proper software support, MD/RAID-456 has not been very competitive against hardware RAID in enterprise use cases. Particularly, MD/RAID-456 suffers from potential data loss during unexpected shut down (the write hole); and long write latency from read-modify-write updates (slow fsync). With recent development in NVRAM (NVMe SSD, NVDIMM), it is now possible to bring enterprise level reliability and performance to MD/RAID-456. In this presentation, Song Liu will present joint work with Shaohua Li in MD/RAID-456 write journal and write back cache.

avatar for Song Liu

Song Liu

Software Engineer, Facebook
Song Liu is a software engineer in Facebook, working on storage software and hardware design. His interest is in storage and distributed systems.

Wednesday April 20, 2016 3:00pm - 3:50pm PDT
State Ballroom B

4:00pm PDT

NOVA: A Log-structured File System for Hybrid Volatile/Non-volatile Main Memories - Andiry Xu, UC San Diego
Fast non-volatile memories (NVMs) present challenges to
file system designers. Existing file systems built for HDD
and SSD introduce software overheads that would obscure
the performance that NVMs should provide, but proposed
file systems for NVMs either incur similar overheads or fail
to provide the strong consistency guarantees that applications

NOVA is a file system that adapts conventional
LFS techniques to exploit the NVM characteristics
and provides strong consistency guarantee. NOVA maintains
separate logs for each inode to improve concurrency, and
stores file data outside the log to minimize log size and reduce
garbage collection costs. NOVA’s logs provide metadata,
data, and mmap atomicity, and put indexes in DRAM to
accelerate search operations. Experimental results show that
NOVA outperforms existing file systems by a wide margin.


Andiry Xu

UC San Diego
I am a PhD student of Non-Volatile Systems Lab, University of California, San Diego. I'm working under the guidance of Professor Steven Swanson. My research involves operating systems and software optimizations to fully exploit the performance potential of next-generation storage... Read More →

Wednesday April 20, 2016 4:00pm - 4:50pm PDT
State Ballroom B
Thursday, April 21

10:30am PDT

BlueStore: A New, Faster Storage Backend for Ceph - Sage Weil, Red Hat
Traditionally Ceph has made use of local file systems like XFS or btrfs to store its data. However, the mismatch between the OSD's requirements and the POSIX interface provided by kernel file systems has a huge performance cost and requires a lot of complexity. BlueStore, an entirely new OSD storage backend, utilizes block devices directly, doubling performance for most workloads. This talk will cover the motivation a new backend, the design and implementation, the improved performance on HDDs, SSDs, and NVMe, and discuss some of the thornier issues we had to overcome when replacing tried and true kernel file systems with entirely new code running in userspace.

avatar for Sage Weil

Sage Weil

Ceph Project Leader, Red Hat
Sage Weil helped developed Ceph as part of his graduate studies at the University of California, Santa Cruz. After graduating he continued to develop the system and build an open source community around Ceph with the support of DreamHost. In 2012, he co-founded Inktank to productize... Read More →

Thursday April 21, 2016 10:30am - 11:20am PDT
State Ballroom E

10:30am PDT

Persistent Memory and the Handling of Media Errors: How to have your Poison and (not) Consume it too - Vishal Verma, Intel
2016 will be the year of Persistent Memory, and these high speed/capacity, non-volatile memory devices are just around the corner, we’re told. Similar to both current storage and memory devices, NVDIMMs will also be susceptible to developing ‘bad’ locations. Unlike DRAM, these errors will persist across reboots, and have to be handled like we do traditional storage.

With DRAM, if a poison location is read, the CPU takes a machine check exception, and typically reboots, and the poison is cleared. With NVDIMMs, the poison won’t be cleared, and it is likely that the application accessing that location will do so again, causing a reboot loop.

This talk details how we solve this problem and its different manifestations in the Linux kernel, and how we enable userspace applications to become aware of poison blocks and deal with them if they so choose.


Vishal Verma

Vishal is a software engineer working for Intel's Open Source Technology Center (OTC). He works on Linux enabling for upcoming technologies. He has been working on enabling for Persistent memory (NVDIMMs) of late, and has spoken at Vault (2015) and Linux Plumbers' Conference (2015... Read More →

Thursday April 21, 2016 10:30am - 11:20am PDT
State Ballroom A

11:30am PDT

Huge Indexes: Algorithms to Track Objects in Cache Tiers - Dan Lambright, Red Hat
A storage cache must implement an index to quickly locate the objects it holds. The index’s design is impacted by the storage medium. For example, a memory cache’s requirements differ from a cache built using storage tiers. In the former, an in-memory hash table or balanced tree may suffice. But in the later, those structures may stumble. The metadata required to track such a large number of objects won’t fit in memory. In such cases, the challenge is to find an index that scales. A further consideration is wether to track elements in LRU order, in which case a sorting mechanism is called for. This talk contrasts 3 cache tiering implementations in Linux that have tackled this problem from GlusterFS, Ceph, and DMcache. Solutions vary from bloom filters to sqlite databases. We will explore their relative pros and cons along the dimensions of performance, overhead, complexity, and more.

avatar for Dan Lambright

Dan Lambright

Software Engineer, Red Hat
Dan Lambright is a principal software engineer at Red Hat, where he works on distributed storage systems. Prior to Red Hat is worked at EMC, DELL, and several storage startups. He also teaches as an adjunct professor at the University of Massachusetts, Lowell.

Thursday April 21, 2016 11:30am - 12:20pm PDT
State Ballroom A

2:00pm PDT

Burst Buffer High Performance Storage with 2 Tiers, OrangeFS and Object Stores in the Cloud and on Premise - Mike Marshall, Clemson University
Over the past decade object stores have become more and more prevalent. They provide infinitely addressable storage but cannot be accessed by millions of lines of currently written IO code. The 2-Tier project is integrating the concepts of distributed OrangeFS burst buffer storage optimized for flash with the ability to tier data seamlessly to an object store. This talk will give an overview of the project, its status and provide attendees a peek into recent technology demonstrations on premise and in the AWS cloud with CloudyCluster.

avatar for Mike Marshall

Mike Marshall

Mike Marshall was introduced to Unix while a Computer Science student at Clemson University in South Carolina around 1982. Clemson students used an IBM Mainframe at that time, but Mike had a part-time job in the Forestry department where they were using a Radio Shack model 16B running... Read More →

Thursday April 21, 2016 2:00pm - 2:50pm PDT
State Ballroom E

2:00pm PDT

Toward A Unified Block IO Controller - Shaohua Li, Facebook
Linux block cgroup currently has two controllers. blk-throttle is bandwidth/IOPS based. CFQ cgroup is proportion based. Having two separate IO controllers is painful. And CFQ is known not performing optimal for high end storage. Also block multiqueue doesn't support CFQ, leaving block-mq only supports blk-throttle. It's time to have a unified IO controller supporting both bandwidth/IOPS and proportion based control and performing well with high end storage. In this presentation, Shaohua Li will discuss current status of IO controller, challenges of a unified IO controller, proposed solution and development status.


Shaohua Li

Software Engineer, Facebook
Shaohua Li is a software engineer in Facebook, working on storage and performance tuning. He has more than 10 years experience in Linux kernel. His interest is storage, memory management and performance tuning.

Thursday April 21, 2016 2:00pm - 2:50pm PDT
State Ballroom A

4:00pm PDT

Panel Discussion: Large-scale Enterprise Automation of Open Source File Systems at Clemson University
Clemson is migrating from large proprietary file system technologies to open source software. We are mid­deployment of over 10 petabytes of SAS­attached JBOD storage for multiple use cases including traditional file system backup solutions, database backups, user/group file services, and HPC file services.

From the start, deployment and operation of these environments presented challenges; from hardware decisions to performance profiling/benchmarking, from storage pool creation to day­to­day user and dataset management. We’ll discuss these real world pain points and how we solved them (or plan to solve them). Born of our frustrations, what started out as a few basic administration scripts has become a multifunctional automated administration utility whose purpose is to make large­scale deployment and operation of open technologies realistic and feasible.


Mike Cannon

Clemson University
Mike Cannon has been with Clemson University since December 2006. Prior to Clemson, Mike was the Storage Manager for NASA. Mike is the director of CIS.

Kevin Stone

Technical Lead & Design Architect, Clemson University
Kevin Stone has held a variety of roles combining systems, programming, andnetworking. Prior to coming to Clemson in 2008 he built backend infrastructure for a USregional ISP. He is currently the technical lead and design architect for Clemson’s coreinfrastructure team.

Thursday April 21, 2016 4:00pm - 4:50pm PDT
State Ballroom B
Filter sessions
Apply filters to sessions.