Next-Gen Supercomputers Face Disk Failure

Mon, September 11, 2006

The next generation of supercomputers could be crippled by hard drive failures every few minutes, the U.S. Department of Energy has warned, and so it is funding a Petascale Data Storage Institute to solve the problem.

The Los Alamos Laboratory has commissioned RoadRunner, a 32,000 CPU supercomputer from IBM that will operate at petaflop levels—a sustained speed of 1,000 trillion calculations per second. Put another way, this is a quadrillion, a million billion, operations per second.

Thousands of hard disks will be needed to keep the thousands of CPUs supplied with data. And Garth Gibson, an associate professor of computer science at Carnegie Mellon university, who will lead the new institute, has warned that this system "likely will require up to hundreds of thousands of magnetic hard disks to handle the data required to run simulations, provide checkpoint/restart fault tolerance and store the output of these modeling experiments. With such a large number of components, it is a given that some component will be failing at all times."

Current teraflop-level supercomputers, operating at trillions of operations per second, have disk failures once or twice a day, according to Gary Grider, a co-principal investigator at the Los Alamos National Laboratory. Once supercomputers are built out to the scale of multiple petaflops, he said, the failure rate could jump to once every few minutes.

Storage systems for them will need to tolerate many failures, mask the effects of them, and continue to operate reliably. "It’s beyond daunting," Grider said of the challenge facing the new institute. "Imagine failures every minute or two in your PC, and you’ll have an idea of how a high-performance computer might be crippled." He emphasized: "For simulations of phenomena such as global weather or nuclear stockpile safety, we’re talking about running for months and months and months to get meaningful results."

-Chris Mellor, Techworld.com (London)

Related Links:

Check out our CIO News Alerts and Tech Informer pages for more updated news coverage.

You are here: Technology Topics » News Feature

Most Recent Technology Topics Stories

White Papers »

Has Your Server Infrastructure Kept Pace with End User Demands?

Learn how your answer to this question compares to your peers by taking this quick poll. See how your peers are dealing with the challenge of ensuring a highly capable server infrastructure as technological shifts impact the application server platform.

Enterprise Strategy Group: The Case for a New Data-centric DLP White Paper

With increasing data growth, comes increased need for data security. The existing DLP model, with a focus on compliance/enforcement is not sufficient as the data discovery and classification capabilities are not granular enough. Read this paper to find how you can efficiently and accurately manage your risk by rapidly inventorying and classifying your data and then developing remediation workflows that support business needs.

Data Loss by the Numbers

This paper breaks down attack sources into four categories: external, malicious insiders, accidental insiders, and unknown.

Data Loss Prevention Solution for Managing E-Discovery

The rapid growth of data and technology is creating challenges for organizations as this digital data is considered to be business communications and must be preserved according the same industry-specific regulations governing the retention and discovery of emails and more traditional forms of electronic communications. This paper examines the role that Data Loss Prevention ("DLP") technology can play in helping organizations address the challenges of locating information in response to electronic discovery.

Best Practices in Data Protection - Executive Summary

This research, conducted by the Ponemon Institute, focuses on issues relating to the use of data protection solutions such as endpoint encryption and data loss prevention within the workplace.

Business Centric DLP

This report, by Jon Oltsik from Enterprise Strategy Group, examines the need for a new business-centric approach to DLP in order to align business and security requirements.

Webcasts »

Virtualization Success is a Team Effort

As greater numbers of datacenter servers transition from the physical to the virtual world, the components of virtualization success come to the fore. What scores of organizations have discovered is that success is derived from an optimal pairing of the right software platform with the right hardware platform.

Customer Spotlight: How IPC The Hospitalist Company Implemented Oracle on VMware

Have you been looking to hear about customer's experiences with the new VMware vCenter Site Recovery Manager product? View this webcast to learn about VMware customer, Navicure, and their experiences testing and evaluating the recovery manager, their progress in implementing it in their environment and their advice other customers considering using vCenter.

Quantifying the Business Value of VMware View - Webcast

Many enterprises have discovered that the use of virtualization to support desktop workloads creates a range of significant benefits. These benefits include price efficiencies, improved IT management and greater agility and choice for end users.

This VMware sponsored webcast with IDC will provide both quantitative measurement of the business value -- defined as the expected ROI -- and qualitative analysis associated with the use of VMware View™. IDC will also provide an analysis of the View Composer and ThinApp™ features of VMware View, including the business value of these solutions and an overview of how they work.

Attend this webcast to learn about:
- Challenges and barriers that might impede the adoption of desktop virtualization
- Navigating roadblocks to facilitate a strategic implementation
- Optimizing qualitative and quantitative benefits to IT and your business

Deliver Private Cloud Database as a Service with VMware vFabric Data Director

VMware recently announced VMware vFabric™ Data Director, a new database deployment and operations platform that enables enterprise IT organizations to offer database as a private cloud service. Built on top of VMware vSphere 5, vFabric Data Director enables IT organizations to ontrol database sprawl through automation and consistent policy enforcement and accelerate application development cycles with self-service database management. Attend this webcast to learn how vFabric Data Director can help you build database-as-a-service in your datacenter.

Reliable Disaster Protection with VMware vCenter Site Recovery Manager

A simple, cost-effective disaster-recovery solution for virtual environments is high on the agenda for IT organizations as they virtualize more business-critical applications with VMware. VMware vCenter™ Site Recovery Manager-the market-leading disaster-recovery product-ensures the simplest and most reliable disaster protection for all virtualized applications. VMware vCenter Site Recovery Manager provides centralized management of recovery plans, enables nondisruptive testing and automates site-failover processes.

Introduction to VMware vCenter Site Recovery Manager 5

Traditional disaster recovery solutions are often too expensive, complex and unreliable to meet business requirements. As a result, IT departments are hesitant to expand disaster protection beyond their most critical applications, largely because they are uncertain whether the quality of the protection is really worth its cost. VMware vCenter™ Site Recovery Manager 5 is the market-leading disaster recovery product that addresses this situation for organizations of all kinds. It complements VMware vSphere to ensure the simplest and most reliable disaster protection for all virtualized applications.

Newsletter Sign-Up »

Receive the latest news test, reviews and trends on your favorite technology topics

Choose a newsletter