Skip to main content
IBM Redbooks

IBM Storage Scale System 6000 with NVIDIA SuperPOD Deployment Guide

An IBM Redpaper publication

thumbnail 

Published on 11 March 2025

  1. .PDF (3.7 MB)
Share this page:
 
 

ISBN-10: 0738462039
ISBN-13: 9780738462035
IBM Form #: REDP-5746-00

Authors: Chris Maestas, Ana Gabriela Iturbe Desentis, Phillip Gerrard, Kiran Ghag, Matthew Klos, John Lewars, Jesus Daniel Munoz Lopez, Roger E. Sanders, Sanjay Sudam, Lindsay Todd and Joanna Wong

    menu icon

    Abstract

    The business world is being transformed rapidly by artificial intelligence (AI), high-performance computing (HPC), analytics, and hybrid cloud technologies. Unlike traditional applications that work with structured data that is stored in databases, these new workloads operate on a vast ocean of unstructured data, such as documents, images, videos, audio recordings, log files, or any other information that can be generated by users, applications, or even machines.

    As a result, organizations are having to re-think their data storage strategies and adapt to new ways of working, including how to use AI to unlock the value of their data, wherever it might be. Information technology (IT) leaders face many new challenges:

    - Accessing and analyzing data that is scattered across the globe

    - The increasing time and resources that are needed by AI training and inferencing workloads

    - The cost and scarcity of resources, such as NVIDIA graphics processing units (GPUs)

    Addressing these challenges requires specialized software and hardware:

    - IBM Storage Scale: Software-defined file and object storage for both structured and unstructured data.

    - IBM Storage Scale System 6000: A hardware implementation of IBM Storage Scale software that is optimized for the most demanding AI, HPC, analytics, and hybrid cloud workloads.

    - IBM Storage Scale System 3500: A hardware implementation of IBM Storage Scale software that is designed for organizations requiring enterprise-ready, entry-level, or mid-level storage system.

    This IBM Redbooks® publication is designed to provide information that is related to the implementation and deployment of the NVIDIA DGX SuperPOD architecture by using IBM Storage Scale System hardware. It is suitable for anyone looking to learn more about the overall solution and provides direct examples of solution implementation.

    Table of Contents

    Chapter 1. Introduction and technical overview

    Chapter 2. Architecture

    Chapter 3. Deployment

    Chapter 4. Server tuning

    Appendix A. IBM Storage Scale System 6000 hosts file for NVIDIA SuperPOD

    Appendix B. IBM Storage Scale System NVIDIA SuperPOD Solution Network Installation Worksheet