ExaStor

    ExaStor is high-performance scale-out storage optimized for AI/ML workloads.
    The Luster-based parallel distributed file system is suitable for high-density,
    large-scale storage workloads that require extreme I/O and scalability,
    such as AI/ML and big data analysis.
    Inquiry
    Click

    Challenges for modern high-performance workloads

    The storage bottleneck issue
    Storage can easily become a main cause of bottleneck when it is not prepared for fast growing AI/ML workloads
    Escalating storage costs
    AI/ML workflow requires tremendous amount of storage space and network traffic which leads to more operational costs
    Increasing operational overhead
    To fully support the dynamic needs of AI applications, storage must provide transparent monitoring and insights throughout the pipeline

    Exascale storage for
    high-performance workloads

    High-performance Parallel File System
    Provide high-speed parallel I/O to every data in the storage cluster
    NVIDIA® GPUDirect® Storage Support
    Support NVIDIA GPUDirect Storage for direct memory access between GPU and storage
    Storage Type in All Scenarios

    Leverage NVMe all-flash, SSD and HDD hybrid configuration

    Flexible Deployment and Scalability

    Choose how to deploy or scale your storage that suits your environment

    Up to 40GB/s per Node

    Delivers 40GB/s sequential read throughput with all-NVMe storage system

    RDMA-powered File Service

    Supports NFS over RDMA protocol with InfiniBand EDR and HDR network

    System Architecture

    High-performance distributed
    file storage optimized for AI/ML workloads

    High-performance Parallel File System

    Allows parallel data access through Luster parallel file system, with high-speed file sharing and petabyte-level scalability through its parallel I/O architecture. Files are distributed and stored in objects across the cluster through global namespace.

    NVIDIA® GPUDirect® Storage Support

    Provides a storage interface to accelerate GPU applications, minimizing CPU and memory load. Maximizes GPU performance by providing a distributed file system IO driver module for applications running on GPGPU.

    Flash-based Level 2 Cache

    Can leverage SSD drives as secondary read cache, extending the main memory cache. When the main memory is full, the system can perform read request on the allocated cache drive, instead of using slower hard drives.

    Integrated Lineup

    Integrated Dual-controller Architecture

    • Features dual-controller hardware architecture with integrated management and data nodes
    • Supports high availability with an active-active setup within a single chassis
    • Up to 40GB/s performance per node with NVMe SSDs
    • Start with a single node with the ability to scale in single-node increments

    Integrated Single-controller Architecture

    • Features an integrated management/data node design based on single-controller hardware
    • Supports high-availability with mirrored nodes
    • Can be configured as all-NVMe or SAS hybrid
    • Requires minimum of 1 node, scalable in 1-node increments

    Separated Lineup

    Separated Single-controller Architecture

    • Features a single controller hardware architecture with management node and data node separated
    • Utilizes SAS/SATA interface for logical disk sharing
    • Supports a 2-node high availability setup and a 4-node 3+1 RAID configuration
    • Begins with a 4-node set and can be expanded per set
    Are you interested in our products?
    Get in touch with Gluesys.