Real-time Data Prefetching Algorithm Based on Sequential Patternmining in Cloud Environment

In cloud computing, data access delay becomes the leading factor that affects the QoS. Data prefetching is an effective technology for hiding data access delay in cloud computing environment. This paper proposes a Real-time Data Prefetching algorithm based on sequential pattern mining to hidden the data access delay. First, it uses sequential pattern mining algorithm to extract user access patterns from user´s historical accesses records. Second, according to users´ access patterns predicts related data object of the data object on demand as a candidate prefetching data set. Finally, according to slackness value of related data objects, they are prefetched to local data nodes. Experiments results showed that this algorithm can effectively reduce the network overhead in data transmission and avoids excessive prefetching.