ITO America Business Technology Publication
 
              
About ITO America
IT Optimization Solution Centers
CIO Resource Centers
BSM & ITILv3
BI
Customer Centric Enterprise and Customer Experience Management
Data Center Optimization
Data Center Security and Integrity
Enterprise Mobility
Green IT and Energy Efficient Data Centers
SaaS and Cloud Computing
Service Oriented Architecture and BPM
Unified Business Communications
Virtualization and Data Center Automation
WAN Application Delivery
CIO Thought Leadership and Contributions
 Podcasts
Videos
Webinars
Business Technology Analyst Partners and Media Partners
ITO America Testimonials and Previous Clients
ITO America Media Kit
ITO America Information Technology Optimization News
Contact ITO America staff
ITO America IT Industry Events
ROI Analysis

Business Technology Whitepapers and Articles
Business Technology Article
PDF Document Download

Document:

Optimizing Power Consumption in Large Scale Storage Systems



You must be a logged in to view this document.

Click here if you've lost or would like to change your password?
Register on ITO America Business Technology website

Register on ITOAmerica.com and gain access to countless articles, podcasts, videos and more, contributed by some of the top minds in business technology!


E-mail:
Password:
Verify Password:
Password must be 7-15 characters in length.
Articles in Business Technology
Excerpt:

Data centers are the backend of a large number of services that are taken for granted today. Cornell University's Computer Science Department members examine a new solution to optimize power consumption in large scale storage systems. They argue that technological evolution has given the Log-Structured File System (LFS) new relevance as a natural power-saving opportunity for large-scale storage systems.


LFS proves to be an accessible solution to disk power management when considering the challenges involved, such as:

  • Short idle periods- LFS localizes write-access to a small subset of disks, this feature results in long disk idle periods
  • Low predictability of idle periods- LFS evades this problem by predetermining which disk is written to at all times
  • Performance constraints- LFS provides an application-independent solution that allows the system to perform consistently across a wide range of datasets
  • The law of large numbers- large scale server systems process incredibly large request loads

The Log-Structured File System was inspired by a need to optimize the latency of write-accesses. Ultimately, findings show that a new opportunity for saving power in large-scale storage systems is born. The solution is that log structured file systems write only to the log head; as a result, if read accesses are served by the cache, then write accesses touch only the log head disk, potentially permitting the power down of all other disks.