Multi-Path Data Prefetching in Mobile Cloud Storage

The high latency nature of mobile network limited the use of mobile cloud storage, especially for these applications with frequent random accesses to a large set of small files. Prefetching techniques are usually utilized to overcome the latency problem, but most existing prefetching algorithm cannot prefetch enough data to alleviate the negative effect of high latency in a random access sequence of small files. By analyzing the efficiency of prefetching, and rely on a probability view of prefetching sequence, a new Multi-Path Prefetching approach is proposed. Instead of pursuing high prefetching accuracy, Multi-Path Prefetching tend to prefetch more successors to ensure high prefetching hit rate and efficiency, and achieve a higher overall throughput performance of accessing cloud storage services via mobile network.