A family of codes with a natural two-dimensional structure is presented, inspired by an application of RAID type of architectures whose units are solid state drives (SSDs). Arrays of SSDs behave differently to arrays of hard disk drives (HDDs), since hard errors in sectors are common and traditional RAID approaches (like RAID 5 or RAID 6) may be either insufficient or excessive. An efficient solution to this problem is given by the new codes presented, called partial-MDS (PMDS) codes.

Considerable interest has been paid in recent literature to codes combining local and global properties for erasure correction. Applications are in cloud type of implementations, in which fast recovery of a failed storage device is important, but additional protection is required in order to avoid data loss, and in RAID type of architectures, in which total device failures coexist with silent failures at the page or sector level in each device. Existing solutions to these problems require in general relatively large finite fields. The techniques of integrated interleaved codes (which are closely related to generalised concatenated codes) are proposed to reduce significantly the size of the finite field, and it is shown that when the parameters of these codes are judiciously chosen, they outperform codes optimising the minimum distance with respect to the average number of erasures that the code can correct.

A new class of codes, Extended Product (EPC) Codes, consisting of a product code with a number of extra parities added, is presented and applications for erasure decoding are discussed. An upper bound on the minimum distance of EPC codes is given, as well as constructions meeting the bound for some relevant cases. A special case of EPC codes, Extended Integrated Interleaved (EII) codes, which naturally unify Integrated Interleaved (II) codes and product codes, is defined and studied in detail. It is shown that EII codes often improve the minimum distance of II codes with the same rate, and they enhance the decoding algorithm by allowing decoding on columns as well as on rows. It is also shown that EII codes allow for encoding II codes with an uniform distribution of the parity symbols.

No abstract

In general, array codes consist of m × n arrays and in many cases, the arrays satisfy parity constraints along lines of different slopes (generally with a toroidal topology). Such codes are useful for RAID type of architectures, since they allow to replace finite field operations by XORs. We present expansions to traditional array codes of this type, like Blaum-Roth (BR) and extended EVENODD codes, by adding parity on columns. This vertical parity allows for recovery of one or more symbols in a column locally, i.e., by using the remaining symbols in the column without invoking the rest of the array. Properties and applications of the new codes are discussed, in particular to Locally Recoverable (LRC) codes.