Efficient data storage: adaptively changing chunk size in cloud computing storage systems

Cloud computing enables users to utilise a shared pool of resources on demand. BLOB storage is a type of cloud storage for unstructured data. Most data management systems use chunk sizes equal to a given BLOB. Despite the simplicity of this strategy, as the nodes in the system are heterogeneous, the BLOB sizes are different and access to the data is not consistent due to these issues. We propose an adaptive strategy in order to define an adequate chunk size by taking into account metrics concerning the actual resources (bandwidth, storage usage), the size of the BLOB/file and the access rate of the chunks. Our experimental results show that our proposed approach provides an execution time that is about 24% better than that of a method based only on one chunk size and about 96% better than that of a method based on a random choice of the chunk size.