CARC web archive · Archived copy of ahpcc.unm.edu/projects/vista-azul/proposal/node11.html as captured by the Internet Archive on 22 February 2001. Links to pages that were not saved here go to the Wayback Machine. Icons the Internet Archive did not save were redrawn for this copy and are not originals (hover over one to see). Wayback copy · All pages in this snapshot · Web archive index
next (modern recreation, not an archived original) up (modern recreation, not an archived original) previous (modern recreation, not an archived original)
Next: I/O Performance Up: Data Intensive Applications Research Previous: Goals And Objectives

I/O Workload Characteristics

What are the local and global I/O workload characteristics of numerically and data intensive applications in terms of temporal and spacial distribution of I/O operations and the amount of data transferred per operation? What are the significant differences between the I/O workloads for numerically and data intensive applications? How do they scale? Here local I/O is defined as I/O WITHIN a node and between a node and its local storage, while global I/O is defined as the I/O BETWEEN nodes of a system and between nodes and main storage. What are the advantages of each of the architectures (clustered SMP nodes and clustered Netfinity nodes) when required to effectively support fast data movement? Does the high performance switch technology hide or amplify the difference in node technology?


Bob Ballance
1999-12-07