AUTHOREA
Log in Sign Up Browse Preprints

loading page

It’s Time to Talk About HPC Storage: Perspectives on the Past and Future
  • +1
  • Bradley Settlemyer,
  • George Amvrosiadis,
  • Philip Carns,
  • Robert Ross
Bradley Settlemyer
Los Alamos National Laboratory
George Amvrosiadis
Carnegie Mellon University
Philip Carns
Argonne National Laboratory
Robert Ross
Argonne National Laboratory

Corresponding Author:[email protected]

Author Profile

Abstract

High-performance computing (HPC) storage systems are a key component of the success of HPC to date. Recently, we have seen major developments in storage-related technologies, as well as changes to how HPC platforms are used, especially in relation to artificial intelligence and experimental data analysis workloads. These developments merit a revisit of HPC storage system architectural designs. In this paper we discuss the drivers, identify key challenges to status quo posed by these developments, and discuss directions future research might take to unlock the potential of new technologies for the breadth of HPC applications.

Preprint status:Published

26 Aug 2021Submitted to Computing in Science and Engineering
Show details
Hide details
27 Aug 2021Editor invited a reviewer
27 Aug 2021Reviewer #1 has been removed by the editor
27 Aug 2021Editor invited a reviewer
28 Aug 2021Review Report #2 received
29 Sep 2021Published in Computing in Science and Engineering
01 Nov 2021Published in Computing in Science & Engineering volume 23 issue 6 on pages 63-68. 10.1109/MCSE.2021.3117353
10.22541/au.163111064.40864768/v1
Kathryn Mohror posted a review
This is a review of the article "It’s Time to Talk About HPC Storage: Perspectives on the Past and Future," for the Leadership Computing department.Comments:Great article! I have only a few small suggestions for clarity/readability below. Abstract: suggest changing "especially as relates to the use of HPC" to "especially related to the use of HPC"paragraph 2: " ORNL Health Data Sciences Institute" -> "Oak Ridge National Laboratory (ORNL) Health Data Sciences Institute"Fig 2 caption: "Each access must traverse multiple distinct protocol hops, with its own interrupt…