PROJECT TITLE :

Enabling Efficient and Reliable Transition from Replication to Erasure Coding for Clustered File Systems - 2017

ABSTRACT:

To balance performance and storage potency, fashionable clustered file systems typically 1st store data with replication, followed by encoding the replicated data with erasure coding. We have a tendency to argue that the commonly used random replication does not take under consideration erasure coding in its style, thereby raising both performance and availability problems in the following encoding operation. We tend to propose encoding-aware replication, which fastidiously places the replicas thus on (i) eliminate cross-rack downloads of data blocks during the encoding operation, (ii) preserve availability while not knowledge relocation once the encoding operation, and (iii) maintain load balancing across replicas as in random replication before the encoding operation. We have a tendency to conduct extensive HDFS-based mostly testbed experiments and discrete-event simulations, and demonstrate the performance gains of encoding-aware replication over random replication.