Evaluation of the Effect of Hypervisor and Container Type Virtualization on Different Workloads Performance in Cloud Environments

Abstract

Advances in virtualization technologies that provide efficient use of physical resources have played major role in development of technologies such as cloud computing, internet of things and software defined networks. Today, container technologies have emerged as alternative to hypervisor virtualization solutions. Cloud providers design a system that supports both hypervisor and container virtualization solutions to provide users with more efficient environments. In this study, performance evaluation of processor, memory, network and disk intensive workloads on KVM hypervisor and LXD container virtualization solutions in OpenStack cloud infrastructure has been performed by applying ‘Domain Knowledge-based Methodology’. Performance tests were performed with PerfKit Benchmarker and Cloudbench benchmark automation tools by using Iperf as a network-intensive workload, HPL as a processor-intensive workload, Stream as a memory-intensive workload and Fio as a disk-intensive workload in the OpenStack cloud environment. As a result, LXD virtualization in OpenStack infrastructure performed better than KVM in processor-intensive, network-intensive, sequential read bandwidth, sequential read input/output per second, sequential write bandwidth and sequential write input/output per second for HDD workloads. KVM virtualization in OpenStack infrastructure performed better than LXD in memory-intensive and random read bandwidth, random read input/output per second, random write bandwidth and random write input/output per second for HDD workloads.

Keywords

• Virtualization
• cloud computing
• hypervisor
• container
• KVM
• LXD

Bulut Ortamlarında Hipervizör ve Konteyner Tipi Sanallaştırmanın Farklı Özellikte İş Yüklerinin Performansına Etkisinin Değerlendirilmesi

Abstract

Fiziksel kaynakların verimli kullanılabilmesini sağlayan sanallaştırma teknolojilerindeki ilerlemeler, bulut bilişim, nesnelerin interneti ve yazılım tanımlı ağ teknolojilerinin gelişiminde büyük pay sahibi olmuştur. Günümüzde hipervizör sanallaştırma çözümlerine alternatif olarak konteyner teknolojileri ortaya çıkmıştır. Bulut sağlayıcıları kullanıcılarına daha verimli ortamlar sunmak için hem hipervizör hem konteyner sanallaştırma çözümlerini destekleyen sistemleri tercih etmektedirler. Bu çalışmada, ‘Alan bilgisine dayalı’ metodoloji uygulanarak OpenStack bulut altyapısında işlemci, bellek, ağ ve disk yoğunluklu iş yüklerinin KVM hipervizör ve LXD konteyner sanallaştırma çözümleri üzerinde performans değerlendirmesi yapılmıştır. OpenStack bulut ortamında PerfKit Benchmarker ve Cloudbench kıyaslama otomasyon araçları vasıtasıyla ağ yoğunluklu iş yükü olarak Iperf, işlemci yoğunluklu iş yükü olarak HPL, bellek yoğunluklu iş yükü olarak Stream ve disk yoğunluklu iş yükü olarak Fio kıyaslama araçları kullanılarak performans testleri gerçekleştirilmiştir. Yapılan testler sonucunda OpenStack bulut altyapısında LXD sanallaştırma KVM sanallaştırmaya göre işlemci, ağ, sabit disk sürücüsünde sıralı okuma bant genişliği, saniyedeki giriş/çıkış sayısı ve sıralı yazma bant genişliği, saniyedeki giriş/çıkış sayısı iş yüklerinde daha iyi performans sergilemiştir. KVM sanallaştırma ise LXD sanallaştırmaya göre bellek, sabit disk sürücüsünde rasgele okuma bant genişliği, saniyedeki giriş/çıkış sayısında ve rasgele yazma bant genişliği, saniyedeki giriş/çıkış sayısında daha iyi performans sergilemiştir.

Keywords

• Sanallaştırma
• bulut bilişim
• hipervizör
• konteyner
• KVM
• LXD

References

1. 1. Awasthi, S., Pathak, A. ve Kapoor, L. (2016) Openstack-paradigm shift to open source cloud computing & its integration, 2016 2nd International Conference on Contemporary Computing and Informatics (IC3I), Noida, 112-119. doi:10.1109/IC3I.2016.7917944
2. 2. Barik, R. K., Lenka, R. K., Rao , K. R. ve Ghose, D. (2016) Performance analysis of virtual machines and containers in cloud computing, 2016 International Conference on Computing, Communication and Automation (ICCCA), Noida, 1204–1210. doi:10.1109/CCAA.2016.7813925
3. 3. Casalicchio, E. ve Perciballi, V. (2017) Measuring Docker Performance: What a mess!!!, ICPE '17 Companion: Proceedings of the 8th ACM/SPEC on International Conference on Performance Engineering Companion, L'Aquila, Italy, 11-16. doi:10.1145/3053600.3053605
4. 4. DeMuro, J. (2018). Konu: What is container technology?. Erişim Adresi: https://www.techradar.com/news/what-is-container-technology (Erişim Tarihi: 17.05.2019).
5. 5. Feitelson, D. G. (2006) Experimental Computer Science:The Need for a Cultural Change, Academic Paper, School of Computer Science and Engineering The Hebrew University of Jerusalem, Israel
6. 6. Felter, W., Ferreira, A., Rajamony, R. ve Rubio, J. (2015) An updated performance comparison of virtual machines and Linux containers, 2015 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS), Philadelphia, PA, 171-172. doi:10.1109/ISPASS.2015.7095802
7. 7. Graber, S. (2016). Konu: LXD 2.0: Introduction to LXD [1/12]. Erişim Adresi: https://stgraber.org/2016/03/11/lxd-2-0-introduction-to-lxd-112/ (Erişim Tarihi: 17.05.2019).
8. 8. Gupta, S. ve Gera, D. (2016) A Comparison of LXD, Docker and Virtual Machine, International Journal of Scientific & Engineering Research, 7(9), 1414-1417.

9. 9. IBM Developer, (2018). Konu: Cloudbench (CBTOOL). Erişim Adresi: https://developer.ibm.com/open/projects/cloudbench-cbtool/ (Erişim Tarihi: 17.05.2019).
10. 10. Kang, M., Kang, D., Walters, J. P. ve Crago, S. P. (2017) A Comparison of System Performance on a Private OpenStack Cloud and Amazon EC2, 2017 IEEE 10th International Conference on Cloud Computing (CLOUD), Honolulu, CA, 310–317. doi:10.1109/CLOUD.2017.47
11. 11. Kozhirbayev, Z. ve Sinnott, R. O. (2017) A performance comparison of container-based technologies for the Cloud, Future Generation Computer Systems, 68, 175-182. doi: 10.1016/j.future.2016.08.025
12. 12. Li, Z., O’Brien, L., Cai, R. ve Zhang, H. (2012) Towards a Taxonomy of Performance Evaluation of Commercial Cloud Services, 2012 IEEE Fifth International Conference on Cloud Computing, Honolulu, HI , 344–351. doi:10.1109/CLOUD.2012.74 Uludağ Üniversitesi Mühendislik Fakültesi Dergisi, Cilt 25, Sayı 2, 2020
13. 13. Li, Z., O’Brien, L. ve Kihl, M. (2016) DoKnowMe: Towards a Domain Knowledge-driven Methodology for Performance Evaluation, ACM SIGMETRICS Performance Evaluation Review, 43(4), 23–32. doi: 10.1145/2897356.2897360
14. 14. Linux Containers, (2019). Konu: LXC Introduction. Erişim Adresi: https://linuxcontainers.org/lxc/introduction/ (Erişim Tarihi: 17.05.2019).
15. 15. Mell, P. ve Grance, T. (2011) The NIST Definition of Cloud Computing, National Institute of Standards and Technology Information Technology Laboratory, doi:10.6028/NIST.SP.800-145
16. 16. Morabito, R., Kjällman, J. ve Komu, M. (2015) Hypervisors vs. lightweight virtualization: A performance comparison, 2015 IEEE International Conference on Cloud Engineering, Tempe, AZ, 386-393. doi:10.1109/IC2E.2015.74
17. 17. OpenStack Docs, (2019). Konu: Welcome to OpenStack Documentation. Erişim Adresi: https://docs.openstack.org/pike/ (Erişim Tarihi: 17.05.2019).
18. 18. PerfKitBenchmarker, (2015). Konu: PerfKit Benchmarker. Erişim Adresi: http://googlecloudplatform.github.io/PerfKitBenchmarker/ (Erişim Tarihi: 17.05.2019).
19. 19. Popek, G. J. ve Goldberg, R. P. (1974) Formal requirements for virtualizable third generation architectures, Communications of the ACM, 17(7), 412–421. doi:10.1145/361011.361073
20. 20. Qian, L., Luo, Z., Du, Y. ve Guo, L. (2009) Cloud Computing: An Overview, IEEE International Conference on Cloud Computing, Springer, Berlin, Heidelberg, 5931, 626–631. doi:10.1007/978-3-642-10665-1_63
21. 21. Ruan, B., Huang, H., Wu, S. ve Jin, H. (2016) A Performance Study of Containers in Cloud Environment, Asia-Pacific Services Computing Conference, Springer, Cham, 10065, 343-356. doi:10.1007/978-3-319-49178-3_27
22. 22. Sampathkumar, A. (2013). TigerPrints Virtualizing Intelligent River®: A Comparative Study of Alternative Virtualization Technologies, M.Sc. thesis, Clemson University, Computer Science, Clemson, South Carolina, United States
23. 23. Wang, B., Song, Y., Cui, X. ve Cao, J. (2017) Performance comparison between hypervisorand container-based virtualizations for cloud users, 2017 4th International Conference on Systems and Informatics (ICSAI), Hangzhou, 684–689. doi: 10.1109/ICSAI.2017.8248375
24. 24. Yemelianov, A. (2017). Konu: Containerization Mechanisms: Cgroups. Erişim Adresi: https://blog.selectel.com/containerization-mechanisms-cgroups/ (Erişim Tarihi: 17.05.2019).