Data caching in multi-container systems
https://doi.org/10.15514/ISPRAS-2019-31(6)-7
Abstract
Today, virtualization is a key technology for cloud computing and modern data centers, providing scalability and security, managing the global IT infrastructure and reducing costs. Among the methods of virtualization, the most popular was containerization, that is the isolation of related groups of linux processes that share a common Linux kernel. Containerization is more profitable that classical hardware virtualization because of compactness of containers and lower overhead costs of memory, disk, CPU. However, in comparison with classical architectures without process isolation containers can cost more, and in any case, the industry is waiting for additional optimization – the speed of launch, saving memory and disk space and other resources. Different caching techniques can help in this, because Caching is the oldest mechanism of increasing software productivity without radical modification of algorithms and hardware. However, there are a lot of architectural and engineering tradeoffs. Here we will consider modern scientific and technical approaches to their solution in different aspects – acceleration of launch, optimization of shared usage, acceleration of building container images, as well as some security problems caused by aggressive caching in modern processor architectures. And in some use cases for multi-container systems performance and latency are not important, but we have to ensure the maximum load of physical servers. In these cases, the algorithms of planning and placement of containers are relevant, and we give an overview of theoretical work on this topic.
Supporting agencies: The studies, the results of which are presented in this article, were supported by the Russian Foundation for Basic Research (project 17-07-01006).
About the Authors
Dmitry Andreevich Grushin
Ivannikov Institute for System Programming of the Russian Academy of Sciences
Russian Federation
Researcher
Russian Federation
Researcher
Denis Olegovich Lazarev
Ivannikov Institute for System Programming of the Russian Academy of Sciences,
Moscow Institute of Physics and Technology (State University)
Russian Federation
Research trainee at ISP RAS and a postgraduate student of MIPT
Russian Federation
Research trainee at ISP RAS and a postgraduate student of MIPT
Stanislav Aleksandrovich Fomin
Ivannikov Institute for System Programming of the Russian Academy of Sciences,
Moscow Institute of Physics and Technology (State University)
Russian Federation
ISPRAS programmer, lecturer at MIPT
Russian Federation
ISPRAS programmer, lecturer at MIPT
References
1. Latency numbers every programmer should know. URL https://gist.github.com/hellerbarde/2843375, accessed 15.11.2019.
2. Увеличение плотности контейнеров на ноде с помощью технологии PFCACHE. Блог компании Rusonyx / Increase container density on a node using PFCACHE technology. Rusonyx company blog. URL https://habr.com/ru/company/rusonyx/blog/444696/, accessed 15.11.2019.
3. Virtuozzo User's Guide. URL https://docs.virtuozzo.com/pdf/virtuozzo_7_users_guide.pdf, accessed 15.12.2019.
4. Павел Емельянов. CRIU – как маленький open-source проект меняет жизнь большой компании. Материалы 11-й международная конференция Linux Vacation/Eastern Europe, 2015, http://0x1.tv/20150627H / Pavel Emelyanov. CRIU – how a small open-source project changes the life of a large company. In Proc. of the 11th International Linux Vacation / Eastern Europe Conference, 2015, http://0x1.tv/20150627H (in Russian).
5. Николай Ефанов. Единая теория восстановления деревьев процессов Linux – огибая подводные камни checkpoint-restore. Материалы 15-й Центрально-восточноевропейской конференция в России по разработке программного обеспечения, 2019, http://0x1.tv/20191115BN / Nikolay Efanov. The unified theory of Linux process tree restoration - enveloping the pitfalls of checkpoint-restore. In Proc. of the 15th Central & Eastern European Software Engineering Conference in Russia, 2019, http://0x1.tv/20191115BN (in Russian).
6. Mike Rapoport, Adrian Reber. To Kill or to Checkpoint – That is the Question. In Proc. of the Open Source Summit + Embedded Linux Conference & OpenIoT Summit Europe, 2018. https://static.sched.com/hosted_files/osseu18/2a/kill-or-checkpoint.pdf.
7. Grushin D.A., Kuzyurin N.N. On Effective Scheduling in Computing Clusters. Programming and Computer Software, vol. 45, issue 7, 2019, pp. 398–404. doi:10.1134/S0361768819070077.
8. Moritz Lipp, Michael Schwarz et al. Meltdown: Reading Kernel Memory from User Space. In Proc. of the 27th USENIX Security Symposium, 2018, pp. 973-990.
9. Paul Kocher, Jann Horn et al. Spectre Attacks: Exploiting Speculative Execution. In Proc. of the IEEE Symposium on Security and Privacy, 2019, pp. 1-19.
10. Paul Turner. Retpoline: a software construct for preventing branch-target-injection. URL https://support.google.com/faqs/answer/7625886, accessed 15.12.2019.
11. Николай Пасынков. Слои Docker для ускорения сборки проекта. Материалы 15-й Центрально-восточноевропейской конференция в России по разработке программного обеспечения, 2019, http://0x1.tv/20191114DJ / Nikolay Pasynkov. Docker layers to speed up project builds. In Proc. of the 15th Central & Eastern European Software Engineering Conference in Russia, 2019, http://0x1.tv/20191114DJ (in Russian).
12. Report to Congress on Server and Data Center Energy Efficiency: Public Law 109-431. U.S. Department of Energy, Office of Scientific and Technical Information. URL https://www.osti.gov/servlets/purl/929723, accessed 15.12.2019.
13. Dow Eli M. Decomposed Multi-Objective Bin-Packing for Virtual Machine Consolidation. PeerJ Computer Science, vol. 2, 2016, article e47.
14. Xia Binzhou, Zhiyi Tan. Tighter Bounds of the First Fit Algorithm for the Bin-Packing Problem. Discrete Applied Mathematics, vol. 158, issue 15, 2010, pp. 1668-1675.
15. György Dósa. The Tight Bound of First Fit Decreasing Bin-Packing Algorithm Is FFD(I) ≤ 11/9OPT(I) + 6/9. Combinatorics, Algorithms, Probabilistic and Experimental Methodologies. Lecture Notes in Computer Science, vol. 4614, 2007, pp. 1-11/
16. Roytman Alan, Aman Kansal et al. PACMan: Performance aware virtual machine consolidation. In Proc. of the 10th International Conference on Autonomic Computing, 2013, pp. 83-94.
17. Beloglazov Anton, and Rajkumar Buyya. Optimal Online Deterministic Algorithms and Adaptive Heuristics for Energy and Performance Efficient Dynamic Consolidation of Virtual Machines in Cloud Data Centers. Concurrency and Computation: Practice and Experience, vol. 24, issue 3, 2012, pp. 1393-1550.
18. A Simple Proof of the Inequality FFD (L) ≤ 11/9 OPT (L) 1, ∀L for the FFD Bin-Packing Algorithm. Acta Mathematicae Applicatae Sinica, vol. 7, issue 4, 1991, pp, 321–331.
19. Piraghaj, Sareh Fotuhi, Amir Vahid Dastjerdi, Rodrigo N. Calheiros, and Rajkumar Buyya. A Framework and Algorithm for Energy Efficient Container Consolidation in Cloud Data Centers. In Proc. of the IEEE International Conference on Data Science and Data Intensive Systems, 2015, pp. 368 - 375.
20. Ali Qasim. Scaling web 2.0 applications using docker containers on vsphere 6.0. VMware VROOM! Blog. URL https://blogs.vmware.com/performance/2015/04/scaling-web-2-0-applications-using-docker-containers-vsphere-6-0.html, accessed 15.11.2019.
Review
For citations:
Grushin D.A., Lazarev D.O., Fomin S.A. Data caching in multi-container systems. Proceedings of the Institute for System Programming of the RAS (Proceedings of ISP RAS). 2019;31(6):125-144. (In Russ.) https://doi.org/10.15514/ISPRAS-2019-31(6)-7
Email this article 