3rd International Workshop on the Lustre Ecosystem:
Support for Deepening Memory and Storage Hierarchy

Hanover, Maryland
July 25-26, 2017

The Lustre parallel file system has been widely adopted by the high-performance computing (HPC) systems as an effective mechanism for managing large-scale storage and I/O resources. Lustre is an open-source parallel file system technology and heavily used by the world's fastest HPC systems. Lustre achieves unprecedented aggregate performance by parallelizing I/O over file system clients and storage targets at extreme scales. Large-scale checkpoint storage and retrieval, which is characterized by bursty I/O from coordinated parallel clients, has been the primary driver of Lustre development over the last decade.

With the introduction of non-volatile storage technologies, many HPC centers are seeing a proliferation of I/O layers in the end-to-end storage hierarchy that place new demands on Lustre. Effectively managing the node-local memory and these new layers is a new challenge for Lustre and requires new technologies and data management policies to be developed to effectively handle data storage and movement across the I/O stack.

In July of 2017, the 3rd International Workshop on the Lustre Ecosystem will be held in Hanover, Maryland. This workshop series is intended to help explore improvements in the performance, flexibility, and usability of Lustre for supporting diverse application workloads and diverse HPC architectures. The past workshops have helped culminate a discussion on the open challenges associated with enhancing Lustre for diverse applications and architectures, the technological advances necessary, and the associated impacts to the Lustre ecosystem. The 3rd International Lustre Ecosystems Workshop will present a series of invited talks from industry, academia, and US National Laboratories focusing on:

  • Lustre Node-Local Memory Management
  • Multilayered Lustre Storage Architectures
  • Data Flow in Lustre across Multiple I/O Stacks
  • Data Management and Handling in Lustre across Multiple I/O Stacks
  • Data Resiliency and Replication Mechanisms in Lustre across Multiple I/O Stacks
  • Data Provenance in Lustre across Multiple I/O Stacks