Distributed cloud applications acquire and generate tremendous amounts of stored information. The data associated with such applications are distributed widely in geographical terms. Replication services are designed to facilitate and support such cloud applications. A replication service aims to replicate data and select the most appropriate data replica from those available in order to minimize application access time. Current replica services can also exhibit an increase in data acquisition by clients resulting in slower data access. This paper proposes a Partial File Cache (PFC) strategy that minimizes the volume of data transfers under a limited cache space and provides faster data access simultaneously. Traditional caching approaches work well on networks with high latency, however, they can be expensive if an application only requires access to a small part of a large file. To improve cache utilization, our solution makes use of different techniques when caching, such as caching sections or parts of files. The system identifies user hotspots so that fit clients are selected for partial data caching to assist third party cloud storage sites. Results show significant improvement over traditional caching techniques when implemented in geographically spatial cloud environments.
展开▼
