Analyzing the download time of availability codes
By Swanand Ravindra Kadhe, Emina Soljanin, A. Sprintson. 2015
37 citations recorded by the source corpus.
In distributed storage systems, a special sub-class of locally repairable codes, referred to as availability codes, has been proposed to enable recovery of each data block from one of its repair groups. A repair group typically contains a small number of nodes and does not overlap with any other repair group of the same data block. Availability codes have several important benefits, including high degree of fault tolerance, efficient recovery from failures, and efficient access to data by multiple users. In this paper, we study the availability codes from a queuing-theoretical perspective. Specifically, we analyze the average time necessary to download a block of data under the Poisson request arrival model in two service/scheduling scenarios. We compare the availability codes with several alternatives such as MDS codes and replication schemes. Our results indicate that availability codes can minimize the download time in some settings, but are not always optimal.
Subfields: Computer Networks and Communications
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Networks and Communications
- Dimitris Papailiopoulos
- Salim el Rouayheb
- Ramtin Pedarsani
- Rawad Bitar
- Parimal Parag
- Kannan Ramchandran
- Kangwook Lee
- Maximilian Lam
Citations in the corpus
- Speeding Up Distributed Machine Learning Using Codes (2015, background): “In [42], [43], Kadhe and Soljanin analyze the latency performance of availability codes, a class of storage codes designed for enhanced availability.”
- Minimizing latency for secure distributed computing (2017, methodology): “[14] proposed the use of availability codes instead of MDS codes to account for straggler mitigation.”
- Minimizing Latency for Secure Coded Computing Using Secret Sharing via Staircase Codes (2018, background): “The early body of work focused on minimizing latency of content download in distributed storage systems, see e.g., [14], [32]–[34] and later the focus has shifted to using codes for straggler mitigation in distributed computing.”