Large volumes of data, which are being collected for the purpose of knowledge extraction, have to be reliably and efficiently stored. Furthermore, retrieval and processing of large data files have to be fast and efficient. Recent work has proposed using erasure codes to address both goals. This paper is concerned with two problems that involve coding and are of particular importance in practice. The first problem deals with coded distributed sparse matrix multiplication. It is shown how coded computing methods previously proposed for arbitrary matrices can be modified and efficiently used for sparse matrices that occur more often in applications. The second problem is concerned with service rates in coded storage systems. It is shown how judicious combining of coding and replication can be used to shape service rate regions of distributed storage systems.