Summary: WAS or Windows Azure Storage is a cloud storage system that uses erasure coding to maintain the data quality and to keep the cost of the data to a limit. The research in the paper is focused on introducing a new erasure code named as Local Reconstruction Codes (LRC). The main idea behind this code is to lower the amount of code fragments reconstructed online. LRC will reduce the bandwidth in handling repair reads. LRC is incorporated in this research to decrease the overhead in Windows Azure Storage.
Strengths: The use of LRC has the potential to decrease the data transfer pressure on the data network by reducing the overhead. This phenomenon also has a positive impact on reducing the time that is required for reconstruction of data fragments. LRC can make the storage more productive when the overhead is reduced. LRC maintains the quality of storage systems that keeps 3 copies of the data.
Weaknesses: Because LRC cannot tolerate arbitrary 4 failures because is not Maximum Distance Separable. LRC does not provide fault tolerance guarantee as compared to Stepped Combination Codes due to the fact that the length of LRC codes is greater than Stepped Combination Codes. Due to this reason, LRC also does not offer efficient reconstruction.
Questions: How is reading from more fragments compared to LRC which reads from fewer fragments going to affect reconstruction performance?
What is the difference between the response of large scale and small scale cloud object stores to LRC?