Cover Image
市場調查報告書

大規模運算的記憶體和儲存的道路的新規則

New Rules of the Road for Memory and Storage in Large-Scale Computing

出版商 IDC 商品編碼 345745
出版日期 內容資訊 英文 14 Pages
訂單完成後即時交付
價格
Back to Top
大規模運算的記憶體和儲存的道路的新規則 New Rules of the Road for Memory and Storage in Large-Scale Computing
出版日期: 2015年11月27日 內容資訊: 英文 14 Pages
簡介

本報告提供HPC問題的中心、分散式記憶體/儲存架構所扮演的角色的變化相關驗證,現狀與未來發展預測相關的系統性資訊。

IDC的見解

調查概要

概況

  • 中心、分散式儲存架構定義
  • 中心型和分散式記憶體/資料庫方法比較

未來展望

  • 分散式儲存調查:主要的使用案例的要點
  • 調查結果:演算法
  • 調查結果:應用
  • 調查結果:架構

主要的建議

參考資料

  • 相關調查
  • 摘要
目錄
Product Code: US40586615

This IDC study is an assessment of the changing role of central and distributed memory/storage architectures for HPC problems. There are a range of new high-performance data analytics (HPDA) algorithms and applications that are different from traditional modeling and simulation HPC counterparts, and they will ultimately drive development of new hardware, software, and systems architectures. The demands for new HPDA visualization capabilities will grow as HPDA applications increasingly call for new ways to better display data output to help subject-matter experts conduct deeper and more substantive analysis. Finally, the need for effective workflow management tools will become even more acute as more HPDA systems will be required to pull double duty as an HPDA batch system for large static data set jobs and a transactional system that ingests both static and live data sets as well as continual user input, perhaps in a 24-hour-a-day uptime environment.

"IDC sees the role of storage as a distinct architectural building block likely undergoing significant changes in reaction to emerging and evolving HPDA requirements. Ultimately, the distinction between computational and storage servers may dissolve as each loses its individual mission and becomes more blurred in an overall scheme of data management and processing." - Bob Sorensen, research vice president, Technical Computing.

Table of Contents

1. IDC Opinion

2. In This Study

3. Situation Overview

  • Defining Centralized and Distributed Storage Architectures
    • Table: Matrix of Storage and Memory Options
  • Comparing Centrally Located and Distributed Memory/Database Approaches
    • Table: Pros and Cons of Shared System Architectures (Quadrants 1 and 2)
    • Table: Pros and Cons of Distributed System Architectures (Quadrants 3 and 4)
    • Table: Best Use Cases for Shared and Distributed System Architectures

4. Future Outlook

  • Distributed Storage Study: Key Use-Case Takeaways
    • Storage Architecture Will Become Increasingly Use Case Specific
    • Storage Schemes Will Continually Evolve to Meet New Requirements
    • Visualization Requirements Will Present New and Interesting Complexities
    • Workflow Is an Increasingly Critical Consideration
    • HPDA Applications Are Different from Traditional HPC Counterparts
  • Phase 2 IDC Findings: Algorithms
    • HPC Algorithms Are Mature; HPDA Algorithms Are in Their Infancy
    • The Commercial HPDA World Is Divided into Two Camps
  • Phase 2 IDC Findings: Applications
    • Transactional Analysis Comes to the Fore
    • The HPDA World Moves at a Fast Pace
    • Benchmarks Will Matter More than Ever
  • Phase 2 IDC Findings: Architectures
    • HPDA Won't Greatly Alter HPC Architecture in the Short Term, But It Will Change Configurations
    • Commercial Machine Learning Users Find Cloud-Based Services an Attractive Option
    • The Storage Model for HPDA Is Unresolved
    • Centralized Resource Management Is Needed for Advanced Analytics
    • Cloud Computing Will Experience Steady Growth and Influence in HPC

5. Essential Guidance

6. Learn More

  • Related Research
  • Synopsis
Back to Top