In this experiment we rerun the same experiments that we have done in experiment
0
46
1 and experiment 2 but under a slower network environment. The delays between different connections remain the same but bandwidths are adjusted. Bandwidth between two transit nodes is reduced to 500Mbps. Bandwidth between transit node and sub node
is reduced to 50Mbps. Bandwidth between two stubs nodes is reduced to 5Mbps.
Bandwidth Transit-Transit 500 Mbps
Bandwidth Transit-Sub 50 Mbps
Bandwidth Sub-Sub 5 Mbps
Table 6 Network setting of experiment 3
Rerun of Experiment 1
Figure 4 the average download times of experiment 1 rerun
From figure 4 we can see the performance of different download methods are actually very similar. The result strengthens the conclusion we made in the previous
0
47
experiment: that when network utilization is high, the average download times of peers is not affected by the way they choose to download the files.
Aside from the conclusion we already know, there is something interesting we observed from our results. Even though average download times across different download methods are similar, but there are notable differences between the
performance between λ=30 and λ= 100 under low bandwidth environment. The result is very easy to understand because we limit our network bandwidth capacities. So, under a low bandwidth environment with shorter average entry time, the upload bandwidth of peers provided reached the limit capacities of backbone network, resulting longer average download times.
Rerun of Experiment 2
Just like the previous rerun of experiment 1, we observed the same result as experiment 2 at this rerun. Different download methods actually do not influence the average download times. We also find the trend that when the average arrival times are lower the average download times are shorter. This is due to the fact that faster arrival of peers may cause backbone network capacities be occupied.
48 Figure 5 the average download times of experiment 2 rerun
In this chapter we mainly run three sets of experiments. The first one is to compare the influence of different download approaches to the average download time. The second one is to examine the situation in which sizes of files are not unique when using MFMT download approaches. The last one is a rerun of previous two experiments under a low bandwidth environment. The result showed no matter under high or low
bandwidth conditions, the download approaches used or the size difference between download files does not affect the average download speed. We also find that when network bandwidth capacities are low, entry rates actually may affect the download performance, because too many peers may occupy the bandwidth, causing longer average download times.
49
Chapter 5 Conclusions
In this paper, we first introduced the various different methods of multiple-file downloads in BitTorrent. Then, we classified and introduced the differences between these approaches. After, we tried to study the performance of different multiple-file download approaches under steady-state based on the previous research from others and some new models of our own.
Unfortunately, our attempt to model the downloading time of MFMT download failed. We tried to study the performance of MFMT with simulation experiments. The results show that MFMT with Inter-torrent collaboration, MFMT without Inter-torrent collaboration and MFST have similar performances.
We also extended the number of unchoking slots in MFST when there are multiple files in the single torrent. The results show the same as other previous research on a single file system, that the number of unchoking slots does not affect the download
50
performance under the steady-state.
Future improvement
We tried to use the commonly used Fluid model to analyze the different approaches of multiple-file download in BitTorrent. Unfortunately, even with so much abstraction of network details, we still failed to find a meaningful result to describe the performance of MFMT. We hope in the future we can develop a new model that is able to better describe the performance of multiple-file download in BitTorrent under various Network conditions.
We have run a series of experiments by using GPS simulators. Even though simulation is a good way to capture the performances of BitTorrent under certain network conditions, we can still consider doing simulation in a real-world network environment. Or, we can even try to collect data and tracker trace from real-world BitTorrent clients and tracker, resulting in a more diverse and complete research on multiple-file download in BitTorrent.
51
Bibliography
[1] BitTorrent. Available: http://www.bittorrent.com/
[2] B. Cohen, "Incentives Build Robustness in BitTorrent," presented at the 1st Workshop on Economics of Peer-to-Peer Systems, 2003.
[3] sandvine, "Global Internet Phenomena Report," Fall 2011.
[4] D. Qiu and R. Srikant, "Modeling and performance analysis of BitTorrent-like peer-to-peer networks," presented at the Proceedings of the 2004 conference on Applications, technologies, architectures, and protocols for computer communications, Portland, Oregon, USA, 2004.
[5] B. Fan, J. C. S. Lui, and D.-M. Chiu, "The design trade-offs of BitTorrent-like file sharing protocols," IEEE/ACM Trans. Netw., vol. 17, pp. 365-376, 2009.
[6] A. R. Bharambe, C. Herley, and V. N. Padmanabhan, "Analyzing and Improving a BitTorrent Networks Performance Mechanisms," in INFOCOM 2006. 25th IEEE International Conference on Computer Communications. Proceedings, 2006, pp.
1-12.
52
[7] L. Guo, S. Chen, Z. Xiao, E. Tan, X. Ding, and X. Zhang, "Measurements, analysis, and modeling of BitTorrent-like systems," presented at the Proceedings of the 5th ACM SIGCOMM conference on Internet Measurement, Berkeley, CA, 2005.
[8] T. Ye, W. Di, and N. Kam-Wing, "Analyzing Multiple File Downloading in
BitTorrent," in Parallel Processing, 2006. ICPP 2006. International Conference on, 2006, pp. 297-306.
[9] L. Guo, S. Chen, Z. Xiao, E. Tan, X. Ding, and X. Zhang, "A performance study of BitTorrent-like peer-to-peer systems," Selected Areas in Communications, IEEE Journal on, vol. 25, pp. 155-169, 2007.
[10] F. Bin, J. C. S. Lui, and C. Dah-Ming, "The Design Trade-Offs of BitTorrent-Like File Sharing Protocols," Networking, IEEE/ACM Transactions on, vol. 17, pp. 365-376, 2009.
[11] N. Laoutaris, D. Carra, and P. Michiardi, "Uplink allocation beyond choke/unchoke: or how to divide and conquer best," presented at the Proceedings of the 2008 ACM CoNEXT Conference, Madrid, Spain, 2008.
[12] M. Izal, G. Urvoy-Keller, E. W. Biersack, P. Felber, A. A. Hamra, and L.
Garces-Erice, "Dissecting BitTorrent: Five Months in a Torrent's Lifetime,"
presented at the PAM, 2004.
[13] W. Yang and N. Abu-Ghazaleh, "GPS: a general peer-to-peer simulator and its
53
use for modeling BitTorrent," in Modeling, Analysis, and Simulation of Computer and Telecommunication Systems, 2005. 13th IEEE International Symposium on,
2005, pp. 425-432.
[14] Bittorrent Protocol Specification v1.0. Available:
http://wiki.theory.org/BitTorrentSpecification
[15] The Pirate Bay. Available: http://thepiratebay.org/
[16] W.-C. Liao, F. Papadopoulos, and K. Psounis, "Performance analysis of
BitTorrent-like systems with heterogeneous users," Perform. Eval., vol. 64, pp.
876-891, 2007.
[17] A. L. H. Chow, L. Golubchik, and V. Misra, "BitTorrent: An Extensible Heterogeneous Model," in INFOCOM 2009, IEEE, 2009, pp. 585-593.
[18] A. Legout, G. Urvoy-Keller, and P. Michiardi, "Rarest first and choke algorithms are enough," presented at the Proceedings of the 6th ACM SIGCOMM
conference on Internet measurement, Rio de Janeriro, Brazil, 2006.
[19] J. A. Pouwelse, P. Garbacki, D. H. J. Epema, and H. J. Sips, "The Bittorrent P2P File-Sharing System: Measurements and Analysis," presented at the 4TH INTERNATIONAL WORKSHOP ON PEER-TO-PEER SYSTEMS (IPTPS), 2005.
[20] Z. Chao, P. Dhungel, W. Di, L. Zhengye, and K. W. Ross, "BitTorrent Darknets," in INFOCOM, 2010 Proceedings IEEE, 2010, pp. 1-9.
54
[21] G. Neglia, G. Reina, Z. Honggang, D. Towsley, A. Venkataramani, and J. Danaher,
"Availability in BitTorrent Systems," in INFOCOM 2007. 26th IEEE International Conference on Computer Communications. IEEE, 2007, pp. 2216-2224.
[22] Y. Yang, A. L. H. Chow, and L. Golubchik, "Multi-torrent: a performance study and applications," Int. J. Adv. Media Commun., vol. 4, pp. 31-58, 2010.
[23] J. Han, T. Chung, S. Kim, H. Kim, T. T. Kwon, and Y. Choi, "An Empirical Study on Content Bundling in BitTorrent Swarming System," CoRR, vol. abs/1008.2574, 2010.
[24] D. S. Menasche, A. A. A. Rocha, B. Li, D. Towsley, and A. Venkataramani,
"Content availability and bundling in swarming systems," presented at the Proceedings of the 5th international conference on Emerging networking experiments and technologies, Rome, Italy, 2009.
[25] E. W. Zegura, K. L. Calvert, and S. Bhattacharjee, "How to model an
internetwork," in INFOCOM '96. Fifteenth Annual Joint Conference of the IEEE Computer Societies. Networking the Next Generation. Proceedings IEEE, 1996, pp. 594-602 vol.2.