Method and System for Background Replication of Data Objects

Background Prefetching is used to improve the speed of data requests by anticipating the next request and prefetching the respective data. Currently prefetching is hand-tuned, which risks being too aggressive or too timid, consequently hurting demand traffic or diminishing prefetching performance. Moreover, hand-tuning requires extra administration costs. Furthermore, most background transfer rates are set by magic numbers to control aggressiveness; however, this method lacks the real-time responsiveness.

Invention Description This system provides a self-tuning methodology to balance prefetch requests against the demand requests across a network. It monitors network conditions on a packet-by-packet basis and reacts to changes in network conditions to guarantee minimal interference while providing excellent performance.

Benefits

Enables system to aggressively prefetch or replicate data without risk of diminishing overall performance Enables easy exploitation of spare bandwidth to gain advantages of background replication Reduces risks of being too aggressive in prefetching

Features

Simplifies applications Real-time responsiveness to cope with changing network conditions

Market Potential/Applications Applications include: Content distribution Application distribution Web prefetching Network file system backup

UT Researcher Michael D. Dahlin, Ph.D., Computer Sciences, The University of Texas at Austin Arunkumar Venkataramani, Ph.D., Computer Sciences, The University of Texas at Austin Ravindranath Kokku, Computer Sciences, The University of Texas at Austin Praveen Yalagandula, MSE, Center for Information Security, The University of Texas at Austin

Type of Offer: Licensing



Next Patent »
« More Computer Science Patents

Share on      


CrowdSell Your Patent