Experts on Data
 
 

Persistent Memory Summit 2018 Presentation Abstracts

2018 Persistent Memory Summit Presentations

Keynote: Realizing the Next Generation of Exabyte-scale Persistent Memory Centric Architectures and Memory Fabrics

Zvonimir Bandic, Senior Director, Research and Development Engineering, Western Digital Corporation

Abstract

In the last five years, the increasing volume, velocity and variety of data generated and consumed by Big Data and Fast Data applications has driven an aggressive pursuit for the next generation of emerging non-volatile memories, particularly in the area of persistent memory. At component level, this memory must be byte-addressable and non-volatile, deliver latency comparable to DRAM, but have density and cost that falls somewhere between DRAM and NAND flash.

Much has been debated about would it take to scale a system to exabyte main memory with the right levels of latencies to address the world’s growing and diverse data needs. This presentation will explore legacy distributed system architectures based on traditional CPU and peripheral attachment of persistent memory, scaled out through the use of RDMA networking. It will discuss the present boundaries of memory and compute technologies, and the many considerations for developing persistent memory, including performance, power, latency requirements and cost merits of parallel and serial attachment points for memories, and show the experimentally measured latency of RDMA access to persistent memory devices.

This presentation will also consider a theoretical question of what would it take to scale a system to exabyte main memory from the perspective of networking fabric required to access such large amounts of main memory at useful latencies. It will explore the “exabyte challenge” from the hardware architecture perspective and, given the present boundaries of memory and compute technologies, quantitatively evaluate latency requirements for memory and memory fabric switch devices. In addition, it will address the ramifications of the large memory footprint of persistent memory for emerging data-intensive workloads, such as high performance data analytics, autonomous vehicles, social networking value extraction, and many traditional memory bound workloads. Finally, it will outline a vision for a prototyping platform for accelerating innovation in networking protocols that will enable experimental evaluation of novel memory fabrics at scale.

Persistent Memory Programming: The Current State and Future Direction

Andy Rudoff, Member, SNIA NVM Programming Technical Work Group and Persistent Memory SW Architect, Intel Corporation

Abstract

Andy will describe the current state of persistent memory programming in the SNIA NVM Programming Model, major operating systems, programming environments, and application software. Andy will also cover work in progress around persistent memory, which promises to produce additional interesting changes to the ecosystem.

File Systems

Amit Golander, Technical Director, NetApp
Neal Christiansen, Principal Development Lead, Microsoft

Abstract

Persistent Memory (PM)-based file systems are key to bringing the low-latency value of PM hardware to applications and businesses. But developing a portable POSIX-compliant file system using the traditional Kernel-to-user space FUSE bridge results in high-latency access, and thus defeats the purpose. To bridge this gap NetApp has developed the open source ZUFS bridge. ZUFS was designed with PM latencies and byte addressability in mind. It enabled NetApp PM-based solution to be more portable and is allowing the open source community to develop a new breed of efficient PM file systems in user space.

Persistent Memory over Fabrics (PMoF)

Paul Grun, Open Fabrics Alliance Vice Chair; Advanced Technology Development, Cray, Inc
Stephen Bates, Chief Technology Officer, Eideticom
Rob Davis, Vice President of Storage Technology, Mellanox Technologies, Inc.

Abstract

It is becoming increasingly clear that the value of Persistent Memory is magnified dramatically if it can be accessed by a client over a fabric. Because of its unique nature, (occupying a unique strata in the classical memory hierarchy), Persistent Memory over Fabrics (PMoF) poses some interesting challenges for the fabric. This session sets the stage for addressing those challenges. The talk begins with a description of three abstract use cases for PMoF as a way of illuminating some of the key fabric requirements necessary to implement it. The next portion describes some of the hardware challenges involved in hosting a remote Persistent Memory service on a switched fabric. The third portion of the talk points toward efforts underway in the industry to upgrade existing fabric technologies to deliver the capabilities needed to support Persistent Memory.

Windows OS Support for PM

Tom Talpey, Architect, Microsoft

Abstract

Pending

Linux Persistent Memory Support - "Ask Me Anything"

Dan Williams, Open Source Developer, Intel Corporation

Abstract

It has been more than 2 years since the first early version of the "pmem" driver was merged in the Linux kernel. Since then, that single standalone driver has grown into family of drivers, and the enabling has branched out to touch the core kernel, memory management, and filesystems. This session provides an overview of the areas of the Linux kernel that have been impacted by persistent memory support, as well as the software component architecture of the operating system support. With that overview as a framework the session opens up for "Ask Me Anything" questions in the realm of Linux kernel persistent memory support.

VMware vSphere Virtualization of PMEM

Richard A. Brunner, Chief Platform Architect & Principal Engineer, VMware

Abstract

Byte-addressable non-volatile memory (PMEM) offers persistence at near DRAM memory speeds in large enough capacity to change the storage vs memory paradigm used by databases, analytics, and filesystems. We believe virtualization is fundamental to workload agility and consolidation for these new applications. We will provide an overview of VMware virtualization for PMEM that is now running on real PMEM products.

Hype to Reality, a Panel Discussion with Leaders of Real World Persistent Memory Applications

Moderator: Jack Vargas, Product Marketing Manager, Intel Corporation
Panel: Scott Miller, Senior Technologist, DreamWorks Animation; Nikita Ivanov, Founder & CTO GridGain Systems; Kodi Umamageswaran, Vice President of Development, Exadata Storage Server and Database Machine, Oracle; Brian Bulkowski, Founder and Chief Technology Officer, Aerospike

Abstract

There are many technologies that are hyped. Four thought leaders from across the industry will sit down and have a discussion on persistent memory. Come to this session and hear from DreamWorks Animation, GridGain, AeroSpike, and Oracle about their journey of turning hype into reality.

Quick and Painless System Performance Acceleration Using NVDIMMs

Brett Williams, Business Development Manager, Micron Semiconductor
Per Brashers,  Storage Strategist, Yttibrium

Abstract

Learn how frequently used data and files can be moved automatically and dynamically to the highest storage tier (Persistent Memory) to accelerate system performance. See analytics that show how specific system/workloads can benefit from the addition of NVDIMMs.

Workload Performance Comparison

Eden Kim, Chief Executive Officer, Calypso Systems Inc.

Abstract

Datacenter Real World Workloads are very different from synthetic lab workloads and can be used to compare performance of various types of storage. SQL Server workloads for a Retail Web Portal and a Navigation Portal are examined and used as test stimulus to compare datacenter storage at the block IO level. A comparison is provided for Persistent Memory and various classes of solid state storage.

Future Media

Bill Gervasi, Principal Systems Architect, Nantero
Andrew Walker, Vice President of Product, Spin Transfer Technologies

Abstract

Carbon nanotube memory (NRAM) promises to be that disruptive memory you’ve dreamed about: DRAM class performance in a non-volatile memory. Across all applications, from SSD through NVDIMM, and from consumer and IoT through enterprise level architectures, replacing not only all volatile memory but also the energy store subsystems of supercapacitors and batteries, NRAM is the inflection point for high performance, low power, low weight, ultra-small solutions.

After 50 years from the inventions of SRAM and DRAM and 30 from that of NAND Flash, the semiconductor memory industry is in the throes of disruption. NAND has already made the transition to 3-D while DRAM approaches its scaling brick wall. Attempts are being made to fill the latency/persistence/cost “gap” in the memory hierarchy between DRAM and NAND with new types of solid-state memories.

This presentation will explain how MRAM will migrate from its existing emerging niche of embedded nonvolatile memory to form a fundamentally important technology for Persistent Memory not only in the “gap” but also as SRAM replacement. The keys to this migration involve both important circuit/system design and magnetic tunnel junction (MTJ) innovations without which MRAM would languish in its existing memory niche.

New Interconnects

Moderator: Doug Voigt, Distinguished Technologist, Hewlett Packard Enterprise
Panel: Mike Krause, Hewlett Packard Enterprise Fellow HPE Enterprise Group; Jeff Suecheli, IBM; David Koenen, David J. Koenen, ARM

Abstract

In October 2016 several new consortia were formed to advance system interconnects between memory and processors, accelerators and IO devices. These interconnects, CCIX, Open CAPI and Gen-Z, are all potentially relevant to the way systems integrate persistent memory. All have made significant progress in the last year. During this session presenters from ARM, IBM and HPE will provide perspective on the roles of these interconnects relative to PM and a status update on significant milestones for the consortia.

Analysts Weigh In on Persistent Memory

Moderator: Michael Oros, Executive Director, SNIA
Panel: Jim Handy, Objective Analysis; Tom Coughlin, Coughlin Associates; Randy Kerns, Enterprise Management Associates; Gil Russell and Alan Niebel, WebFeet Research

Abstract

Pending