The growing adoption of prediction serving systems (PSSes) has made cost-saving deployment on preemptible spot instances crucial, yet frequent preemptions severely harm availability. While coded computation (CC) can keep availability cost-effectively by encoding original jobs into parity ones, its direct application to spot instances incurs prohibitive decoding overhead and tail latency under frequent preemptions. We identify two findings for optimization: (i) decoding asymmetry (only original job failures require decoding); (ii) preemption unevenness (variation in preemption rates across cloud regions). Leveraging these findings, we propose SpotCC, a new CC framework that strategically dispatches parity jobs to high-preemption (volatile) regions and original jobs to low-preemption (stable) regions. SpotCC designs locality-based and fine-grained volatility identification to reduce decoding operations and mitigate job congestion, respectively, and adaptively tunes configurations for decoding minimization. Experiments show that SpotCC improves P99 latency by 83.9% over state-of-the-arts, while maintaining ultra-low monetary costs.

Minimum-storage regenerating (MSR) codes are provably optimal erasure codes that minimize the repair bandwidth (i.e., the amount of traffic being transferred during a repair operation), while minimizing storage redundancy, in distributed storage systems. However, the practical repair performance of MSR codes still has significant room for improvements, as their mathematical structure makes repair operations difficult to parallelize. In this article, we present HyperParaRC, a parallel repair framework for MSR codes. HyperParaRC leverages the sub-packetization nature of MSR codes to parallelize the repair of sub-blocks and balance repair load (i.e., the amount of traffic sent or received by a node) across available nodes. We first demonstrate that there exists a trade-off between repair bandwidth and maximum repair load. We then propose an affinity-based heuristic for HyperParaRC, which approximately minimizes the maximum repair load by examining the bandwidth incurred during sub-block computations and significantly reduces the search time for large coding parameters compared with our earlier work, ParaRC. Based on our affinity-based heuristic, we further design a full-node recovery mechanism for HyperParaRC that combines both intra-stripe and inter-stripe parallel repair scheduling to repair multiple lost blocks in a failed node. We prototype HyperParaRC on Hadoop HDFS and evaluate it on Alibaba Cloud. Our evaluation results show that HyperParaRC reduces both single-block repair and full-node recovery times compared with state-of-the-art repair approaches.

Erasure coding is a known redundancy technique that has been popularly deployed in modern storage systems to protect against failures. By introducing a small portion of coded redundancy into data storage, erasure coding is shown to provide higher reliability guarantees than replication under the same storage overhead. Despite its storage efficiency, erasure coding incurs high-performance overhead in repair and updates, and its reliability also depends on the amount of redundancy. How to resolve the tensions among storage efficiency, performance, and reliability has been the major research direction in the literature for decades. In this article, we present an in-depth survey of the past, present, and future of erasure coding in storage systems. We conduct our survey from a systems perspective, with an emphasis on how erasure coding is deployed in practical storage systems. Specifically, we first review the use of erasure coding in storage systems from both academia and industry and state the challenges of deploying erasure coding in practice. We then review the topics of erasure coding in three aspects: (i) new erasure code constructions, (ii) algorithmic techniques for efficient erasure coding operations, and (iii) erasure coding for emerging architectures. Finally, we provide future research directions for erasure coding.

Redundancy transitioning enables erasure-coded storage to adapt to varying performance and reliability requirements by re-encoding data with new coding parameters on-the-fly. Existing studies focus on bandwidth-driven redundancy transitioning that reduces the transitioning bandwidth across storage nodes, yet the actual redundancy transitioning performance remains bottlenecked by the most loaded node. We present BART, a load-balanced redundancy transitioning scheme that aims to reduce the redundancy transitioning time via carefully scheduled parallelization. We show that finding an optimal load-balanced solution is difficult due to the large solution space. Given this challenge, BART decomposes the redundancy transitioning problem into multiple sub-problems and solves the sub-problems via efficient heuristics. We evaluate BART using both simulations for large-scale storage and HDFS prototype experiments on Alibaba Cloud. We show that BART significantly reduces the redundancy transitioning time compared with the bandwidth-driven approach.

Modern storage systems not only introduce data redundancy for fault tolerance, but also conduct regular main- tenance operations on storage nodes for system robustness. Erasure coding provides storage-efficient redundancy and has been widely deployed in production, yet it also incurs substantial bandwidth and I/O overhead due to the repair of storage failures. In particular, maintenance operations make storage nodes temporarily unavailable and lead to data unavailability, thereby incurring repair overhead for erasure-coded storage. In this paper, we study Locally Repairable Codes (LRCs), a class of practical repair-efficient erasure codes, and show that there exists an inherent performance trade-off between the repair and maintenance operations of LRCs in data center settings, such that the repair performance in regular (i.e., no-maintenance) and maintenance modes cannot be simultaneously optimized. To this end, we design a configurable data placement scheme that operates along the trade-off subject to fault-tolerance constraints. We prototype our data placement scheme atop Hadoop HDFS and show how it balances the performance trade-off of repair and maintenance operations in real network environments.

Erasure coding has been extensively deployed in today's data centers to tackle prevalent failures, yet it is prone to substantial cross-rack traffic for parity updates. In this article, we propose a new rack-coordinated update mechanism to suppress the cross-rack update traffic, which comprises two successive phases: a delta-collecting phase that collects data delta chunks, and another selective parity update phase that renews the parity chunks based on the update pattern and parity layout. We further design${\sf RackCU}$, an optimal rack-coordinated update solution that achieves the theoretical lower bound of the cross-rack update traffic. We also perform reliability analysis, demonstrating that${\sf RackCU}$can attain a lower data loss probability via shortening the update procedure. We conduct extensive evaluations, in terms of large-scale simulation and real-world data center experiments. We show that${\sf RackCU}$can reduce 16.5-77.1% of the cross-rack update traffic and hence improve 24.9-772.0% of the update throughput.

Erasure coding offers a storage-efficient redundancy mechanism for maintaining data availability guarantees in large-scale storage clusters, yet it also incurs high performance overhead in failure repair. Recent developments in accurate disk failure prediction allow soon-to-fail (STF) nodes to be repaired in advance, thereby opening new opportunities for accelerating failure repair in erasure-coded storage. To this end, we present a fast proactive repair solution called${{\sf FastPR}}$, which carefully couples two repair methods, namely migration (i.e., relocating the chunks of an STF node) and reconstruction (i.e., decoding the chunks of an STF node through erasure coding), so as to fully parallelize the repair operation across the storage cluster.${{\sf FastPR}}$solves a bipartite maximum matching problem and schedules both migration and reconstruction in a parallel fashion. We show that${{\sf FastPR}}$significantly reduces the repair time over the baseline repair approaches for both Reed-Solomon codes and Azure's Local Reconstruction Codes via mathematical analysis, large-scale simulation, and Amazon EC2 experiments.

Erasure coding has been extensively deployed in today's data centers to tackle prevalent failures, yet it is prone to give rise to substantial cross-rack traffic for parity update. In this paper, we propose a new rack-coordinated update mechanism to suppress the cross-rack update traffic, which comprises two successive phases: a delta-collecting phase that collects data delta chunks, and another selective parity update phase that renews the parity chunks based on the update pattern and parity layout. We further design RackCU, an optimal rack-coordinated update solution that achieves the theoretical lower bound of the cross-rack update traffic. We finally conduct extensive evaluations, in terms of large-scale simulation and real-world data center experiments, showing that RackCU can reduce 22.1%-75.1% of the cross-rack update traffic and hence improve 34.2%-292.6% of the update throughput.

We propose repair pipelining , a technique that speeds up the repair performance in general erasure-coded storage. By carefully scheduling the repair of failed data in small-size units across storage nodes in a pipelined manner, repair pipelining reduces the single-block repair time to approximately the same as the normal read time for a single block in homogeneous environments. We further design different extensions of repair pipelining algorithms for heterogeneous environments and multi-block repair operations. We implement a repair pipelining prototype, called ECPipe , and integrate it as a middleware system into two versions of Hadoop Distributed File System (HDFS) (namely, HDFS-RAID and HDFS-3) as well as Quantcast File System. Experiments on a local testbed and Amazon EC2 show that repair pipelining significantly improves the performance of degraded reads and full-node recovery over existing repair techniques.

Erasure coding offers a storage-efficient redundancy mechanism for maintaining data availability guarantees in large-scale storage clusters, yet it also incurs high performance overhead in failure repair. Recent developments in accurate disk failure prediction allow soon-to-fail (STF) nodes to be repaired in advance, thereby opening new opportunities for accelerating failure repair in erasure-coded storage. To this end, we present a fast predictive repair solution called FastPR, which carefully couples two repair methods, namely migration (i.e., relocating the chunks of an STF node) and reconstruction (i.e., decoding the chunks of an STF node through erasure coding), so as to fully parallelize the repair operation across the storage cluster. FastPR solves a bipartite maximum matching problem and schedules both migration and reconstruction in a parallel fashion. We show that FastPR significantly reduces the repair time over the baseline repair approaches via mathematical analysis, large-scale simulation, and Amazon EC2 experiments.

Repair performance in hierarchical data centers is often bottlenecked by cross-rack network transfer. Recent theoretical results show that the cross-rack repair traffic can be minimized through repair layering, whose idea is to partition a repair operation into inner-rack and cross-rack layers. However, how repair layering should be implemented and deployed in practice remains an open issue. In this article, we address this issue by proposing a practical repair layering framework called DoubleR . We design two families of practical double regenerating codes (DRC), which not only minimize the cross-rack repair traffic but also have several practical properties that improve state-of-the-art regenerating codes. We implement and deploy DoubleR atop the Hadoop Distributed File System (HDFS) and show that DoubleR maintains the theoretical guarantees of DRC and improves the repair performance of regenerating codes in both node recovery and degraded read operations.