基于镜像层关联的Docker注册表缓存预取策略

doi:10.3778/j.issn.1673-9418.2003025

摘要/Abstract

摘要：

随着容器技术的广泛普及，大型Docker公共注册表使用对象存储服务来解决镜像数量剧增的问题，但这种松耦合的注册表设计导致较高的延迟开销。为了增强注册表性能，提出一种基于镜像层关联的Docker注册表缓存预取策略LCPA，当注册表服务器缓存未命中时，通过分析镜像元数据文件构建镜像的存储结构，由关联度模型对存储结构计算得到相关镜像层集合，并从后端存储中主动式预取回注册表中以提高缓存命中率。经真实工作负载下收集的Docker数据集测试，实验结果表明LCPA策略比LRU、LIRS和GDFS等缓存算法提高12%~29%的平均缓存命中率，拉取镜像的平均延迟节省率提高了21.1%~49.4%。与现有的LPA预取算法相比，拉取镜像的平均缓存命中率提升25.6%。仿真实验表明该策略可以有效地利用缓存空间，大幅提升注册表的缓存命中率，并降低镜像拉取的延迟开销。

关键词: 公共注册表, 关联度模型, 缓存命中率, 主动式预取

Abstract:

With the popularization of container technology, large-scale Docker public registries use object storage services to solve the problem of the sharp increasement in the number of images, but this loosely coupled registry design results in higher latency overhead. A cache prefetching strategy named LCPA (layer correlation prefetch algorithm) is proposed based on the correlation of image layers to enhance registry performance. When the registry server cache misses, LCPA creates the storage structure of layer by analyzing image metadata and calculates it by the correlation model to obtain the set of relevant layers. Finally, the registry actively prefetches the set from the back-end storage to the memory to improve the cache hit ratio. Experiment uses the Docker trace collected in real production workloads to test algorithm. The results show that LCPA outperforms the traditional cache algorithms, such as LRU, LIRS and GDFS etc., improves the average cache hit ratio by 12%-29% and increases the average latency saving by 21.1%-49.4%. Compared with the existing LPA prefetching algorithm, LCPA improves the cache hit ratio by 25.6%. Simulation experiments show that LCPA can effectively utilize the cache, greatly improves the cache hit ratio of Docker registry, and reduces the latency overhead of pulling images.

Key words: public registry, association model, cache hit ratio, actively prefetch

张晨, 邓玉辉. 基于镜像层关联的Docker注册表缓存预取策略[J]. 计算机科学与探索, 2021, 15(2): 249-260.

ZHANG Chen, DENG Yuhui. Cache Prefetching Strategy Based on Correlation of Image Layers in Docker Registry[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(2): 249-260.

参考文献

[1] SOLTESZ S, P?TZL H, FIUCZYNSKI M E, et al. Container-based operating system virtualization: a scalable, high-per-formance alternative to hypervisors[C]//Proceedings of the 2007 EuroSys Conference, Lisbon, Mar 21-23, 2007. New York: ACM, 2007: 275-287.
[2] SCHEEPERS M J. Virtualization and containerization of app-lication infrastructure: a comparison[C]//Proceedings of the 21st Twente Student Conference on IT, Enschede, Jun 23, 2014. Enschede: University of Twente, 2014: 21.
[3] DENG Y. What is the future of disk drives, death or rebirth?[J]. ACM Computing Surveys, 2011, 43(3): 1-27.
[4] DENG Y, WANG F. Exploring the performance impact of stripe size on network attached storage systems[J]. Journal of Sys-tems Architecture, 2008, 54(8): 787-796.
[5] ANWAR A, MOHAMED M, TARASOV V, et al. Improving docker registry design based on production workload ana-lysis[C]//Proceedings of the 16th USENIX Conference on File and Storage Technologies, Oakland, Feb 12-15, 2018. Berkeley: USENIX Association, 2018: 265-278.
[6] HARTER T, SALMON B, LIU R, et al. Slacker: fast distri-bution with lazy docker containers[C]//Proceedings of the 14th USENIX Conference on File and Storage Technologies, Santa Clara, Feb 22-25, 2016. Berkeley: USENIX Associa-tion, 2016: 181-195.
[7] HUANG H, ZHANG H, SHAO D. Impact of automation tools on China??s DevOps practice[J]. Journal of Software, 2019, 30(10): 3056-3070.
黄璜, 张贺, 邵栋. 自动化工具对中国DevOps实践的影响[J]. 软件学报, 2019, 30(10): 3056-3070.
[8] WU S, WANG K, JIN H. Research status and prospect of operating system virtualization[J]. Computer Research and Development, 2019, 56(1): 58-68.
吴松, 王坤, 金海. 操作系统虚拟化的研究现状与展望[J]. 计算机研究与发展, 2019, 56(1): 58-68.
[9] HAO T Y, WU H, WU G Q, et al. Container-level flexible resource supply method for microservice architecture[J]. Com-puter Research and Development, 2017, 54(3): 597-608.
郝庭毅, 吴恒, 吴国全, 等. 面向微服务架构的容器级弹性资源供给方法[J]. 计算机研究与发展, 2017, 54(3): 597-608.
[10] NATHAN S, GHOSH R, MUKHERJEE T, et al. Comicon: a co-operative management system for docker container images[C]//Proceedings of the 2017 IEEE International Con-ference on Cloud Engineering, Vancouver, Apr 4-7, 2017. Washington: IEEE Computer Society, 2017: 116-126.
[11] DU S M, LI C L, MAO X J, et al. The optimization of LRU algorithm based on pre-selection and cache prefetching of files in hybrid cloud[C]//Proceedings of the 17th International Conference on Parallel and Distributed Computing, App-lications and Technologies, Guangzhou, Dec 16-18, 2016. Washington: IEEE Computer Society, 2016: 125-132.
[12] KOTHI N, LAXKAR P, CHAURASIA S, et al. A novel approach to design extended LFU page replacement algori-thm[J]. International Journal of Advanced Science and Tech-nology, 2019, 28(20): 718-726.
[13] JAVAID Q, ZAFAR A, AWAIS M, et al. Cache memory: an analysis on replacement algorithms and optimization techni-ques[J]. Mehran University Research Journal of Engineering & Technology, 2017, 36(4): 831.
[14] ZHAO Z H, MA Y, CONG Q. GDSF-based low access latency Web proxy caching replacement algorithm[C]//Pro-ceedings of the 2018 2nd International Conference on Com-puter Science and Artificial Intelligence, the 10th Interna-tional Conference on Information and Multimedia Techno-logy, Shenzhen, Dec 8-10, 2018. New York: ACM, 2018: 232-236.
[15] CHENG Y, GUPTA A, BUTT A R. An in-memory object caching framework with adaptive load balancing[C]//Procee-dings of the 10th European Conference on Computer Systems, Bordeaux, Apr 21-24, 2015. New York: ACM, 2015: 1-16.
[16] SUN G J, SHEN J J, VEIDENBAUM A V. Combining pre-fetch control and cache partitioning to improve multicore performance[C]//Proceedings of the 2019 IEEE International Parallel and Distributed Processing Symposium, Rio de Janeiro, May 20-24, 2019. Piscataway: IEEE, 2019: 953-962.
[17] LI M, ZHANG H J, WU Y J, et al. Prefetch-aware fingerprint cache management for data deduplication systems[J]. Frontiers of Computer Science, 2019, 13(3): 500-515.