LoneStar RAID

Abstract

The need for huge storage archives rises with the ever growing creation of data. With today’s big data and data analytics applications, some of these huge archives become active in the sense that all stored data can be accessed at any time. Running and evolving these archives is a constant tradeoff between performance, capacity, and price. We present the LoneStar RAID, a disk-based storage architecture, which focuses on high reliability, low energy consumption, and cheap reads. It is designed for MAID systems with up to hundreds of disk drives per server and is optimized for “write once, read sometimes” workloads. We use dedicated data and parity disks, and export the data disks as individually accessible buckets. By intertwining disk groups into a two-dimensional RAID and improving single-disk reliability with intradisk redundancy, the system achieves an elastic fault tolerance that can at least recover all 3-disk failures. Furthermore, we integrate a cache to offload parity updates and a journal to track the RAID’s state. The LoneStar RAID scheme provides a mean time to data loss (MTTDL) that competes with today’s erasure codes and is optimized to require only a minimal set of running disk drives.

Computing the Probability for Data Loss in Two-Dimensional Parity RAIDs

“…Lonestar scheme: A generalisation of the full scheme is the scheme that was developed for the Lonestar archival system [2], [3]. Here the horizontal parity groups do not have to fill one row exactly, but can be shorter.…”

“…DATALOSSPROBLS, DATALOSSPROBSTD and DATALOSSPROBFULL were analysed in [2] and [4]. Using the reliability model introduced in [11], these papers compute the mean time to data loss (MTTDL) of a data storage system which -as the name suggests -is the expected time until disk failures lead to (irreparable) data loss.…”

“…Using the reliability model introduced in [11], these papers compute the mean time to data loss (MTTDL) of a data storage system which -as the name suggests -is the expected time until disk failures lead to (irreparable) data loss. But the analysis in [2] considers only a small number of scenarios with a fixed number of disks, and both studies limit the number of disk failures to 5. We will show in the next section that the generalised problem, while easy for small numbers, is in fact #P-hard for an arbitrary number of disk failures and provide an exponential-time algorithm for DATALOSSPROBSTD and DATALOSSPROBFULL.…”

“…Since more and more data are stored on disks, there is ongoing research on how they can be efficiently secured using RAID schemes and erasure codes, e.g. [2]- [10]. The idea is to prevent data loss due to failing disks by adding redundancy and by automatically restoring the data of failed disks on new ones (forward error correction).…”

“…The solution to this problem is a requirement for computing the overall data loss probability using, for example, the model by Hafner and Rao [11]. This model has been applied in the past [2], [4], [5] to solve the problem for specific RAID schemes, but only for a limited number of disk failures.…”