Locally repairable codes with high availability based on generalised quadrangles (Q2070258)

Locally Repairable Codes (LRC) have been recently introduced to help provide an efficient way to implement the reliability systems used in large distributed storage systems. The RLC approach consists of minimizing the number \(r\) (called \textit{repair degree}) of nodes used to restore the integrity of the data when needed. To further strengthen the reliability of these systems, it is desirable to also maximize the number \(a\) (called \textit{repair availability}) of different \(r\)-subsets of nodes that are available to restore any data symbol. \textit{L. Pamies-Juarez} et al. [``Locally repairable codes with multiple repair alternatives'', in: Proceedings of the IEEE international symposium on information theory, ISIT 2013, Istanbul, Turkey, July 7--12, 2013. Piscataway, NJ: IEEE. 892--896 (2013; \url{doi:10.1109/ISIT.2013.6620355})] used \textit{generalised quadrangles} to construct a class of LRC codes, called \textit{\(pg\)-BLRC codes}. In particular, they showed that if \(G\) is a partial geometry with parameters \((s,t,\alpha)\) and \(C\) is the corresponding binary \(pg\)-BLRC with repair degree \(r\) and repair availability \(a\), then \(r\leq s\) and \(a\geq t+1\). In this paper, the authors determine the explicit values of \(r\) and \(a\) by showing that \(r=s\) and \(a=(p-1)(t+1)\) if \(G\) is a generalised quadrangle and \(C\) is a \(p\)-ary code for some prime \(p\).