MENU

Validation of Lustre ZFS for the HPC Storage Market – Webinar

Lustre delivers an incredible level of performance, flexibility and scalability for better than 60% of the TOP100 supercomputing sites in the world. As the leader in large storage systems, this massively scalable open source file system has become the de facto choice for HPC, Bio-Science, and real-time Big Data analytics.

Intel Lustre ZFS Webinar

The RAID Inc. and Intel joint webinar offers more on Intel Enterprise Edition for Lustre on ZFS.

This tested and proven Lustre platform currently owns a better than 50% share of the HPC storage market. RAID Inc. working as partners with the Intel® Enterprise Edition of Lustre team has successfully deployed Lustre on ZFS solutions with up to 180 GB/s performance. This Intel® supported deployment of Lustre, based on RAID Inc. storage platform can reduce storage costs by better than 50% versus traditional storage.

Intel® Solutions for Lustre software and The Intel® Scalable Systems Framework

Intel HPC scalable system framework enables converged designs that can be optimized for the expected range of workloads of a given installation site. Once built Lustre can even be reconfigured to best serve the changing workloads that run on it from hour to hour.

RAID Inc and Intel Lustre ZFS takes a holistic design approach to overcome  the growing impasses such as the memory, I/O and power walls, as well as the desire to simplify the technology for broader access to new users and via the cloud.

Simply developing a new technology like faster memory is insufficient. Intel has been investing for years to continue to advance the processor, but also

  • Build out the other necessary system technologies
  • Use industry leading design and manufacturing capabilities to integrated in better ways
  • Control Lustre systems with cohesive system software stacks to extract full performance while simplifying the operations

Intel® Solutions for Lustre Software and ZFS overview

Intel Lustre ZFS on Linux is an object based, open source, distributed, parallel, clustered file system (GPLv2) designed for maximum performance at massive scale. Intel EE Lustre is POSIX compliant and includes global, shared name space where all clients can access all data to achieve a resource efficient and cost effective storage at scale solution.

  • Runs externally from compute cluster
  • Accessed by clients over network (Ethernet, InfiniBand)
  • Up to 512 PB file system size, 32 PB per file
  • Production filesystems have exceeded 2TB/sec

Lustre ZFS is functional since Lustre 2.4 with production today at Lawrence Livermore National Labs and many other Lustre sites. Durability of data is key motivation with the use of ZFS on top of Hardware Raid or ZFS RAIDZ — utilizing RAID Inc. Ability JBOD class storage.

Lawrence Livermore National Labs Lustre ZFS on Linux Deployment

The RAID Inc. and Intel webinar covers an architectural overview and technical deep dive into the real-world 17PB (petabyte) deployment at Lawrence Livermore National Labs with its custom-tuned Lustre on ZFS design. Join this webcast to learn Lustre on ZFS deployment tips and tricks in building an efficient and resilient environment at scale.

Speakers

  • Brent Gorda, General Mgr. for HPC Storage in the High Performance Computing Group at Intel
  • Keith Mannthey, Lustre Solutions Architect working in the Intel High Performance Data Division of Intel
  • Brad Merchant, Senior HPC Engineer at RAID Incorporated