1.哈尔滨工程大学智能科学与工程学院,黑龙江 哈尔滨 150001
2.中国电子科技集团第五十八所,江苏 南京 214062
陈 李(1985—),男,博士研究生,E-mail:chenl@cksic.com.
收稿:2026-08-17,
网络首发:2026-10-23,
移动端阅览
陈李,程建华,郭家乐.基于纠删码动态资源感知与修复树并行调度的故障恢复方法[J].华中科技大学学报(自然科学版),
CHEN Li,CHENG Jianhua,GUO Jiale.Fault recovery method based on dynamic resource awareness and repair-tree parallel scheduling for erasure-coded distributed storage[J].J.Huazhong Univ.of Sci.& Tech.(Natural Science Edition),
陈李,程建华,郭家乐.基于纠删码动态资源感知与修复树并行调度的故障恢复方法[J].华中科技大学学报(自然科学版), DOI:10.13245/j.hust.261222.
CHEN Li,CHENG Jianhua,GUO Jiale.Fault recovery method based on dynamic resource awareness and repair-tree parallel scheduling for erasure-coded distributed storage[J].J.Huazhong Univ.of Sci.& Tech.(Natural Science Edition), DOI:10.13245/j.hust.261222.
针对纠删码分布式存储故障恢复中节点资源动态变化导致恢复节点选择不准确、异构链路造成传输路径代价与转发负载不均衡,以及节点独占约束限制多跳任务并行执行等问题,提出一种基于动态资源感知与修复树并行调度的故障恢复方法.该方法通过建立六维节点状态演化模型,采用层次分析法与熵权组合赋权,并结合逼近理想解排序法,构建动态资源感知节点选择(DRRNS)方法,实现恢复节点的动态评估与选择;其次,综合节点评分、链路质量、累计传输代价和转发负载构建时间感知修复树,并采用关键路径优先调度估计理论并行工期,提高节点选择的动态适应性和多跳修复任务的并行效率.仿真实验结果表明:在RS(9,6),RS(12,8)和RS(15,10)三种配置下,与4种端到端组合基线相比,所提方法的平均理论并行工期降低28.06%~60.01%,充分验证了动态节点选择、修复树构建与并行调度的联合优化效果.
To address inaccurate recovery-node selection caused by dynamically varying node resources,path-cost and forwarding-load imbalance over heterogeneous links,and limited parallel execution of multi-hop tasks under node-exclusive constraints in erasure-coded distributed storage recovery,a failure recovery method based on dynamic resource awareness and parallel repair-tree scheduling is proposed.A six-dimensional node-state evolution model is established,and subjective and objective weights are determined by combining the analytic hierarchy process with the entropy-weight method.The Technique for order preference by similarity to ideal solution is then employed to develop a dynamic resource-aware recovery node selection (DRRNS) method,enabling the dynamic evaluation and selection of recovery nodes.A time-aware repair tree is constructed by considering node scores,link quality,cumulative transmission costs,and forwarding loads.Critical-path-priority scheduling is adopted to estimate the theoretical parallel makespan,thereby improving the adaptability of recovery-node selection to resource variations and the parallel efficiency of multi-hop recovery tasks.Simulation results under RS(9,6),RS(12,8),and RS(15,10) configurations show that the proposed method reduces the average theoretical parallel makespan by 28.06%-60.01% compared with four end-to-end combined baselines,demonstrating the effectiveness of jointly optimizing dynamic node selection,repair-tree construction, and parallel scheduling.
Shen Z R , Cai Y H , Cheng K Y , et al . A survey of the past,present,and future of erasure coding for storage systems [J ] . ACM Transactions on Storage , 2025 , 21 ( 1 ): 1 - 39 .
陈李 , 程建华 , 李明明 . 基于TCN-VLSTM网络的全脉冲多雷达信号调制识别方法 [J/OL ] . 系统工程与电子技术 , 2026-09-02 . https://link.cnki.net/urlid/11.2422.TN.20260902.1610.004 https://link.cnki.net/urlid/11.2422.TN.20260902.1610.004 .
邓见光 , 赵跃龙 , 袁华强 . 云计算环境下的动态数据副本管理策略 [J ] . 华中科技大学学报(自然科学版) , 2015 , 43 ( 10 ): 53 - 57 .
Lucas Filho E R , Efstathiou A , Yang L , et al . DITIS:an end-to-end system-level simulator and optimizer for distributed tiered storage [J ] . SN Computer Science , 2025 , 6 ( 6 ): 746 .
Li J , Zhou Z R , Ding J L . Data-loss models for proactive-tolerance Reed-Solomon storage systems [J ] . Future Generation Computer Systems , 2025 , 170 : 107832 .
Chen J M , Li Z P , Fang G , et al . A comprehensive repair scheme for distributed storage systems [J ] . Computer Networks , 2023 , 235 : 109954 .
Chen N J , Liu W T , Pu W J , et al . SDNC-repair:a cooperative data repair strategy based on erasure code for software-defined storage [J ] . Sensors , 2023 , 23 ( 13 ): 5809 .
Xia J X , Luo L L , Sun B W , et al . Parallelized in-network aggregation for failure repair in erasure-coded storage systems [J ] . IEEE/ACM Transactions on Networking , 2024 , 32 ( 4 ): 2888 - 2903 .
Liu L , Wang Y , Liang Y F , et al . In-network aggregation enabled multiple sub-blocks parallel repair in erasure-coded storage system [J ] . Computer Networks , 2025 , 270 : 111523 .
Yang C H , Zhong K , Tan Y J , et al . RAN:accelerating data repair with available nodes in erasure-coded storage [C ] // Proc of 2025 IEEE International Conference on Cluster Computing . Piscataway,NJ : IEEE , 2025 : 1 - 12 .
Liu X Z , Hu Y C , Feng D , et al . Repair friendly wide-stripe erasure coding for in-memory key-value stores [J ] . Journal of Systems Architecture , 2025 , 167 : 103516 .
许亮业 , 石连星 , 单蓉胜 . 一种纠删码条带化数据的一致性检查方法 [J ] . 上海交通大学学报 , 2024 , 58 ( 4 ): 579 - 584 .
邓文杰 , 洪铁原 , 唐聃 , 等 . 一种基于多层次校验的低恢复成本纠删码 [J ] . 计算机应用研究 , 2024 , 41 ( 5 ): 1441 - 1447 .
Zhou Z J , Deng X H , Pei X J , et al . ZJC:constructing fully local repair in erasure codes for distributed cloud storage [J ] . Journal of Systems Architecture , 2026 , 174 : 103735 .
Liu X Z , Hu Y C , Wang W C , et al . Optimizing encoding and repair for wide-stripe minimum bandwidth regenerating codes in in-memory key-value stores [J ] . Journal of Systems Architecture , 2025 , 161 : 103369 .
0
浏览量
0
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621