In order to improve the effect of data rebalancing and reduce the impact of data migration, a context-aware data reba-lancing approach is proposed. First, a predictive model of migration time is presented to depict the impact of virtualization context on data migration. Then a context-aware data rebalancing algorithm (CADR) is provided based on fine-grained resource monitoring. The experiments show that our prediction model of migration time has a low error rate, and CADR can improve perfor-mance compared with a typical data rebalancing algorithm.