Code Conversions in Storage Systems

Distributed storage systems play a pivotal role in reliably storing huge volumes of data across large clusters of storage devices. These systems predominantly employ erasure codes to protect data against device failures in a cost-effective manner. The parameters for these codes are determined based on factors such as failure rate, reliability, performance requirements, and storage overhead. There exist several circumstances that necessitate changing the parameters of an erasure code on already-encoded data. This gives rise to the code conversion problem, wherein already-encoded data using certain parameters must be transformed to conform to a different set of parameters. This article provides an overview of the code conversion problem, discusses existing approaches to code conversion, and outlines promising future research directions. By incorporating this new dimension into the design of erasure codes, storage systems can further optimize resource utilization and reliability, thereby facilitating more efficient data storage at scale.