storage systems

Modern data storage systems suffer from latent cross-layer faults due to tight hardware–software coupling across multiple abstraction layers, often leading to silent data corruption or unrecoverable data loss. To address this, we propose the first cross-layer fault-tolerance analysis framework targeting heterogeneous storage stacks—including SSDs, persistent memory, local file systems, and distributed storage. Our approach combines architectural modeling of the full stack, systematic injection of representative defects, and precise tracking of fault propagation across hardware–firmware–software boundaries to expose error propagation paths and consistency violation mechanisms. Through empirical evaluation across widely deployed systems, we identify critical vulnerabilities impacting data integrity and quantify coverage gaps in existing fault-tolerance techniques. The framework provides a scalable, principled methodology for analyzing cross-layer resilience and establishes concrete, actionable directions for designing next-generation highly reliable storage systems.

This work addresses the joint optimization of media selection, capacity allocation, and data placement (replication vs. tiering) for key-value caching across heterogeneous NVM/DRAM/disk storage under memory budget constraints. We introduce the first systematic modeling framework for multi-level non-volatile cache configurations, analytically characterize the operational regimes where replication or tiering dominates, and propose an adaptive configuration policy grounded in device failure rates and data update frequencies. Our methodology integrates cache access behavior modeling, hierarchical configuration optimization, and empirical validation using memcached benchmarks. Results demonstrate that tiering substantially outperforms replication under low device failure rates and high update workloads. Key contributions include: (1) a deployable, low-overhead configuration algorithm; (2) quantitative design guidelines for heterogeneous cache deployment; and (3) theoretical foundations for the reliability–performance trade-off in tiered caching systems.

To address inefficient data migration and inaccurate performance prediction in heterogeneous storage systems (NVMe cache + HDD backend), this paper designs and implements a distributed two-tier storage system. We propose an online reinforcement learning–based dynamic data tiering scheduling algorithm and develop an end-to-end performance model integrating queuing network theory with fine-grained device behavior modeling. Our key contribution is the first scalable, fine-grained device behavior modeling method tailored for heterogeneous storage—enabling adaptive tiering management and precise performance prediction under high-concurrency I/O workloads in multi-core clusters. Experimental evaluation on multi-node clusters demonstrates an average model prediction error of less than 8%, a 27% improvement in I/O throughput, and a 34% reduction in average access latency. The framework provides a reusable modeling and optimization foundation for two-tier storage systems.

To address the challenge in distributed storage systems of simultaneously achieving low storage overhead, high reliability, and low repair traffic—where replication and erasure coding (EC) individually fall short—this paper proposes HyRES, a network-scale-aware hybrid storage scheme. HyRES innovatively unifies replication and EC within a single coherent framework, rather than merely combining them. It introduces a dynamic tiered encoding strategy and a scale-adaptive repair scheduling mechanism to jointly optimize storage cost, file loss probability (FLP), and cross-network repair traffic. Theoretical modeling and large-scale simulations demonstrate that, under identical fault tolerance guarantees, HyRES reduces storage overhead by approximately 40% compared to pure replication, lowers FLP by over 50% relative to conventional EC, and significantly mitigates the scaling of repair traffic with increasing network size.

Modern storage hierarchies face a fundamental trade-off between load balancing and space efficiency. To address this, we propose Mirror-Optimized Storage Tiering (MOST), a co-design strategy integrating mirroring with tiered storage. MOST implements dynamic hot-data identification and cross-tier mirroring via Cerberus—a user-space storage management layer built atop CacheLib—thereby eliminating the high-overhead data migrations inherent in conventional tiering. Its core innovation lies in employing lightweight mirroring to enhance I/O parallelism and bandwidth utilization while preserving the space efficiency of tiered storage. Experimental evaluation across diverse I/O-intensive and dynamic workloads demonstrates that Cerberus achieves an average 32% throughput improvement over state-of-the-art approaches; gains are especially pronounced in NVMe+SSD hybrid tiers.

Despite widespread adoption, there remains a lack of systematic evaluation on whether cloud Elastic SSDs (ESSDs) can effectively replace local SSDs while delivering comparable performance. Method: This paper presents the first in-depth empirical characterization of ESSDs on AWS and Alibaba Cloud, identifying four counterintuitive behavioral patterns distinguishing them from local SSDs—and introducing the concept of the “implicit contract”: unarticulated but operationally binding design assumptions imposed by cloud providers that fundamentally constrain ESSD performance. Contribution/Results: Based on these findings, we distill five actionable engineering insights to guide storage stack redesign for ESSD-specific behavioral boundaries. We precisely quantify performance inflection points, latency distributions, and resource contention patterns under realistic workloads. Furthermore, we establish a reproducible benchmarking methodology and derive key design principles for cloud-native storage optimization.

To address inefficient data management in heterogeneous storage systems—caused by protocol diversity, fragmented authentication mechanisms, and the absence of a unified coordination framework—this paper proposes the Wide-area Data Distribution System (WDDS). WDDS introduces three key innovations: (1) a “data container” abstraction that unifies interfaces and access semantics across heterogeneous storage sources; (2) an elastic, scalable wide-area storage network integrating erasure coding with dynamic load balancing; and (3) a lightweight distributed authentication model natively supporting S3, POSIX, and WebDAV protocols. Evaluated on medical imaging and satellite remote sensing workloads, WDDS achieves a 10% higher throughput than centralized cloud solutions, faster failure recovery than Redis and IPFS, and sustained service delivery under >10,000 concurrent clients and large-scale node failures.

This study addresses the practical disparities and co-evolution between high-performance computing (HPC) and edge computing architectures within the cloud continuum. It presents the first large-scale empirical analysis based on 396 real-world, production-grade AWS architectures. Methodologically, we propose a multidimensional, data-driven framework encompassing service topology identification, storage type classification, architectural complexity quantification, and ML service integration statistics. Results reveal systematic differences—and complementary patterns—between HPC and edge architectures across four dimensions: core service composition (e.g., EC2 versus Greengrass/Lambda), storage design paradigms (parallel file systems versus distributed lightweight caches), complexity distributions, and ML embedding strategies. This work delivers the first industry-scale architectural benchmark for the cloud continuum, providing empirically grounded insights and methodological foundations for cross-domain architecture design, resource optimization, and cloud-native convergence of HPC and edge computing.

Edge cloud storage faces dual challenges: limited capacity and stringent latency requirements for high-temporal-data access under dynamic workloads, hindering profit maximization. To address this, we propose a profit-driven framework that jointly optimizes dynamic space partitioning and erasure coding. We introduce a novel storage layout dividing edge server capacity into heat-adaptive private and shared public zones. Further, we design an elastic data placement and cache replacement policy driven by request rates, integrated with erasure coding to enable cross-node redundancy and collaborative caching. Extensive experiments on synthetic workloads and real-world traces from Netflix and Spotify demonstrate that our approach improves system profitability by 5%–8% over state-of-the-art methods, while significantly enhancing both operational efficiency and economic viability of edge storage.

Existing storage performance models—such as the Disk Access Model (DAM)—fail to accurately capture the concurrent I/O characteristics of multi-queue SSDs, hindering hardware-aware co-design of external-memory algorithms. Method: We propose MQSSD, a novel storage abstraction model that explicitly incorporates multi-queue parallelism as a fundamental dimension—revealing concurrent access as the key mechanism enabling modern SSDs’ high throughput. MQSSD is derived from joint empirical characterization of real SSD hardware and LSM-tree engines (e.g., RocksDB), combining analytical modeling with system-level validation. Contribution/Results: MQSSD achieves significantly higher prediction accuracy than DAM. Guided by MQSSD, we design an LSM-tree variant optimized for multi-queue SSDs, establishing a new external-memory data structure paradigm centered on “high concurrency and low serial dependency.” This provides a scalable theoretical foundation for hardware-aware algorithm design.