Quick viewing(Text Mode)

A Peep on the Interplays Between Online Video Websites and Online

A Peep on the Interplays Between Online Video Websites and Online

arXiv:1305.4018v1 [cs.SI] 17 May 2013 h aoiyo hs iesaesilhse noln online interestin in an is hosted study it still to Hence, Hulu. topic are and YouTube as Sin such videos. these watch of to majority people for ar the platforms vid largest as online such second (ONSs) the the comScore’ networks besides social to 2013[2], online according February websites, system Recently, in inter-autonomous released video[1]. the report all is of today 25% traffic to 15% that reported nCiacmeigwt ahohr .. uo,Ku6.com, , e.g., other, each with competing in time, same links video. external the this video of At referred the information immediately. have can then summary OSN that he OSN, brief that interesting, an provides to is choose video. shared video button, its be a sharing of will that the each finds on for user click video U.S., Youku function online in YouTube a sharing as popular When famous video a as Youku, such just An is 1. provides blogs. which Fig. China, personal in in and shown ext is Facebook, other example to as sharing such video the websites facilitate to buttons sharing paper. this in motivations our of one forms generated traffic of a 25% flows for analysi video account videos. The of they all c) 10% and by that OSNs. OSNs, illustrates of to data kinds related traffic Weibo two indica real OSNs) the which on of and microblogs), based Chinese Chinese natures Sina popular popular different in most the most Video than two two OSNs. top viewers (the top in more popularity (the attract its usually online on in Renren, itself impact video in greatly a of can factors i popularity websites videos The many of b) are probability websites. there sharing video external a) the that, OSN affect find can websites, that We video traffic. th online network using sources: campus OSNs data and different websites totally sharing video between interplays s and Characterizing ISPs OSNs, them. websites, understand video share for online insights and among websites great relationships provide video watch can videos online to interplays in re-share these this people videos and and more of websites, popularity watch attract video the online may boost from OSNs also friends their webs in by of types shared users behav two example, sharing the between video For interplays Such the espe (OSNs). changes websites networks greatly other social to sharing online video the the facilitate to links Abstract ie rfci ail rwn nteItre.I is It . the in growing rapidly is traffic Video eie ok,teeaemn te nievdowebsites video online other many are there Youku, Besides the people provide websites video online many fact, In the study to experiments empirical conduct we paper this In epo h nepasbtenOln Video Online between Interplays the on Peep A Mn nievdowbie rvd h shortcut the provide websites video online —Many o hs iestase oOSNs to transfer videos these how e stsadOln oilNetworks Social Online and Websites .I I. uzo ho igu ag igTo iooM n Xiaohong and Ma Xiaobo Tao, Jing Wang, Pinghui Zhao, Junzhou NTRODUCTION O LNSLb ia ioogUiest,X’n ..Chin P.R. Xi’an, University, Jiaotong Xi’an Lab, KLINNS MOE Email: { zho hag to ba xhguan xbma, jtao, phwang, jzzhao, hc also which , online n ilyto cially n the ing ,and s, video ernal on. o ites. and can ree ior tes eo ce re g e s s s ecnuta ndphaayi ae ndtst collected sources: datasets data on of based views of analysis perspective depth multiple in from an conduct we their about knowledge realistic ecosystem. online more the a in positions people provide and just are Henc websites. ecosystem situations video to online Their requirements the are than on. there worse even so or to and similar Sin Weibo Qzone, , Tencent Renren, Weibo, e.g., also China, are dec in there unbelievable co-existing websites, these OSNs video to usu- online which confused to China, the Similar person larations. of are third industry they a video that makes online announced ally the them of of bananas Many top on. so and 56.com • • follows: network. as campus findings a our from Weibo. summarize We data Sina traffic and Tudou. Network Renren and OSNs, Youku from websites, Data video 3) online form Data 2) 1) them, between relationships complex the study to order In ouaiyi Ss iesi ernadQzone Tencent and and its Sina Renren in on than in viewers impact itself more Videos greatly attract video OSNs. usually can a in websites of popularity popularity video Sina the online and that in Renren find from we collected Weibo, dataset the is on category, Based its e.g., factor. the video, important a affect an video of also online can video nature in factors The online videos websites. of many from probability that sharing collected external find dataset we the websites, on Based hc om u eodmtvto osuythem study to motivation second our forms which , } @sei.xjtu.edu.cn i.1 ie rfiei Youku. in profile video A 1. Fig. rn rest hsdsree online disordered this to orders bring a Guan many e, a - TABLE I Weibo, which indicates the different natures of the two VIDEO PROFILES IN YOUKU AND TUDOU.A TICK REPRESENTS HAVING kinds of OSNs. THIS ITEM, AND A CROSS REPRESENTS NOT HAVING THIS ITEM. • The analysis based on real traffic data shows that 10% of video flows are associated with OSNs, and they Profile Youku Tudou category √ √ account for 25% of traffic generated by all videos. length (minutes) √ √ size (bytes) √ The outline of this paper is as follows. Section II summa- rating √× × rizes related works. The data collection process is described uploaded date √ √ #views √ √ in Section III. Section IV presents the results of an in-depth #comments √ √ analysis of the collected datasets. Concluding remarks then #favorites √ √ #likes √ √ follow. #dislikes √ √ #external shares √ √ II. RELATED WORK top 20 external shared links √ √ Cha et al. [3] crawled the YouTube and Daum UCC, the most popular UGC service in Korea, and presented an exten- network, which can introduce uncorrectable statistic bias[13], sive analysis of the video popularity distribution, popularity [14]. Hence, results obtained from data collected by BFS evolution, user behavior analysis, and content duplication in is unbelievable. Fortunately, we find that Youku assigns an YouTube. Further, Cheng et al. [4] investigated the social eight digits numeric ID to each videos, so a uniform sample networks in YouTube videos. Chatzopoulou et al. [5] ana- of Youku videos can be obtained by generating uniformly lyzed popularity by looking at properties and patterns in time random numbers, and by polling Youku to known about their and considering various popularity metrics, and studied the existences. Based on this uniform video sampling method, relationship of the popularity metrics. [6] and [7] collected 1.5 millions of videos in Youku are collected. Similar data traces at the edge of a single campus network and studied collection process is also conducted in Tudou, in which we usage patterns, video properties, popularity and referencing also uniformly collected 1.5 millions videos. For each video characteristics, and transfer behaviors of YouTube from the in Youku and Tudou, its profile information is also retrieved, perspective of an edge network. Saxena et al. [8] analyzed which is shown in Table I. and compared the underlying distribution frameworks of three video sharing services—YouTube, and , B. Data Collection from Renren and based on traces collected from measurements performed in a PlanetLab environment. Renren3 is one of the largest OSNs in China with more than 150 million users. Every user can watch or share videos Using traffic trace collected at multiple PoPs of the ISP, in Renren. This sharing can be external (from online video Adhikari et al. [9] inferred the load balancing strategy used by websites to Renren) or internal (inside Renren from a user YouTube to serve user requests. Torres et al. [10] employed to another user). We collect all the Renren accounts related state-of-the-art delay based geolocation techniques to find to Xi’an Jiaotong University (they or their friends are in the geographical location of YouTube servers, and performed the campus), which form a network of 1,661,236 nodes and analysis on groups of related YouTube flows. Compared with 32,050,611 edges. From these users, we further collect 0.5M a location-agnostic algorithm to map users to data centers videos. For each video, we record its original URL link analyzed in [9], they found that the YouTube infrastructure has address, the number of views and shares inside Renren. been completely redesigned and primarily uses a nearest RTT mapping policy. We also collect data from Sina Weibo4, one of the largest microblogs in China, which has more than 200 million users. In [11], [12], the authors studied the impacts of external Sina Weibo assigns a ten digits numeric ID to each user, links on the YouTube and Youku videos. Our work differs itself therefore we can collect tweets from 14,623 uniform randomly by focusing on relationships between online video websites sampled users and about 3.7M tweets are collected. Each tweet and OSNs. And we use data from three totally different sources is classified to a retweet (posted by retweeting an existing to support this study. tweet) or an original tweet (posted by self writing). About 64.8% of the tweets are retweets in our dataset. III. DATA COLLECTION METHODS In this section, we describe the data collection methods C. Traffic Collection from the Edge from three different data sources. The traffic data used in this paper is based on the actual network traffic over the backbones of CERNET (China Edu- A. Data Collection from Youku and Tudou cation and Research Network) Northwest Regional Center and Youku1 and Tudou2 are the top two most popular online the campus network of Xi’an Jiaotong University. The traffic video websites in China. To collect videos in Youku, pre- data is collected at an egress router with a bandwidth of 1.5 vious study[12] uses the simple Breadth-First-Search (BFS) Gbps by using TCPDUMP for about ten days in March 2011. method. However it is known that incomplete BFS is likely to Since application level characteristics are our primary interests densely cover only some specific region of the Youku video in this paper, our analysis focuses on the HTTP traffic data. We

1http://www.youku.com 3http://www.renren.com 2http://www.tudou.com. Tudou was acquired by Youku in March 2012. 4http://www.weibo.com 1.0 107 107 107 6 Youku 6 6 10 Tudou 10 10 0.9 105 105 105 104 104 104 0.8 103 103 103 102 102 102 101 101 101 0 0 0 CDF 0.7 10 10 10 Avg. # external shares 100 101 102 103 104 105 106 107 108 Avg. # external shares 100 101 102 103 104 105 106 Avg. # external shares 100 101 102 103 104 105 #views #comments #favorates 0.6 Youku Tudou 0.5 107 107 107 0 1 2 3 4 5 6 106 106 106 10 10 10 10 10 10 10 105 105 105 Number of external shares 104 104 104 103 103 103 102 102 102 101 101 101 Fig. 2. Cumulative distribution of the number of external shares. 100 100 100 Avg. # external shares 100 101 102 103 104 105 106 107 Avg. # external shares 100 101 102 103 104 105 Avg. # external shares 100 101 102 103 #likes #dislikes Video length (minutes)

107 107 107 group packets into different bidirectional HTTP flows, wherea 106 106 106 105 105 105 flow is defined as a bidirectional, ordered sequence of packets 104 104 104 103 103 103 generated by a pair of (packet source IP, packet source port) 102 102 102 101 101 101 and (packet destination IP, packet destination port). For each 100 100 100 HTTP bidirectional flow, we record its starting time, finishing Avg. # external shares 0 500 1000 1500 2000 Avg. # external shares 0 500 1000 1500 2000 Avg. # external shares 0 1 2 3 4 5 6 Video age (days) Bit rate (kilobytes/minute) Rating time, total bytes, total packets, HTTP request (e.g, the method, URL, Host), and HTTP response (e.g. status code, content Fig. 3. Relationship between video profiles and external shares. type, content size). 40M HTTP flows are collected, which include 13K distinct sources5. •#favorites When a video is interesting to a user, he can mark IV. DATA ANALYSIS it as a favorite video (which results in that the video is saved in his personal homepage in the video site to facilitate his In this section, we analyze the relationships between online future watching). Since marking a video as favorite also needs video websits and OSNs from three different perspective of the login, which may also reduce the correlation with external views. sharing.

A. External Video Sharing in Youku and Tudou •#likes and #dislikes Comparing the two factors, a video receiving more likes will be more likely to be shared than First, we provide overview statistics of the video external a video has less likes or many dislikes. shares in Youku and Tudou. Fig. 2 depicts the empirical cu- mulative distributions (CDF) of the number of external shares •Length Video length has weak correlation with video’s for videos in Youku and Tudou. We find that the majority of external sharing. Comparing short videos with long videos, videos are not widely shared externally. For example, 80% of short videos have higher probability to be shared. This is Youku videos have less than 22 shares, and 80% of Tudou because users may not have enough time to watch a long video videos have less than 18 shares. About 0.3% of the videos in and decide to share it. Youku and Tudou have more than 10,000 external shares, and •Age Age also has weak correlation with a video’s external these minority videos are widely shared outside of Youku and sharing. Videos in Tudou shows that old videos are less likely Tudou. to be shared than new videos. However, this is a little different Second, we want to study what are the factors that may in Youku. cause a video to be widely shared outside of Youku and Tudou? •Bit rate and rating Bit rate of a video is defined as Hence, we consider the video profiles listed in Tab. I. The Size/Length. Both the two factors can be used to characterize relationships between these factors and the average number of the quality of a video. Generally, higher quality videos has external shares are shown in Fig. 3. Most of these factors are higher probability to be shared. However, this relationship is consistent with each other either in Youku or Tudou. We will very weak. analyze them in detail. We use Pearson correlation coefficients (PCC) to quantify •#views The number of views seems to have the strongest these correlations in Youku and Tudou, which are shown in relationship with the possibility of external sharing of a video. Table II. PCC is the most common metric to measure the Intuitively, a video receiving more views indicates itself is dependence between two quantities. For a pair of metrics popular, hence it will be shared to external sites with a high x and y, its corresponding values for the sampled video i probability. And because of the feedback effect, many external (i =1, 2,...,n) are xi and yi, where n is the total number of shares will also increase its views inside of a video site. sampled videos. Then their PCC is defined as •#comments The number of comments has similar effect with n (x − x)(y − y) the number of views, and it also has strong relationship with Pi=1 i i ρxy = n 2 2 the external sharing. However, to comment on a video needs pPi=1(xi − x) P(yi − y) the user to login first, which may stop a lot of people who n n don’t own accounts. where x = Pi=1 xi and y = Pi=1 yi are the sample means of x and y. The results of PCCs are consistent with our analysis 5All recorded IP addresses are anonymized to protect users’ privacy. above. TABLE II TABLE III PCCSOFDIFFERENTFACTORS. TOP EXTERNAL OSNS CONSUMING YOUKU AND TUDOUVIDEOS.

Factors Youku Tudou #external links ( 105) External site #views 0.50 0.54 Youku× Tudou #comments 0.17 0.16 Renren.com 110 220 #favorites 0.14 0.15 Qzone.qq.com 470 120 #likes 0.06 0.05 .com 12 99 #dislikes 0.01 0.04 Douban.com 0.27 1.1 length 0.00 0.00 Sina Weibo (weibo.com) 5.0 4.7 age 0.01 0.00 Tencent Weibo (t.qq.com) 6.4 8.1 bit rate 0.00 - rating - 0.03

6 Youku Tudou 1.0 10 0.93 5 0.32v 4.1 3.6 9.5 8.1 0.9

Others 10 0.6 0.65 4.1 7.1 0.8 4 0 0 0 0 10

Travel 0.6 0.42 2.4 8.3 0.7 3

0.4 1.2 7.4 6.6 10

Fashion CDF 0.6 0.6 0.91 5.1 7.4 2 #shares 10 1.6 2.1 3.2 10.9 0.5 Science

0.9 1.3 4.9 11.8 1 0.4 Shares 10 4.1 2.9 1.8 52.6 Views 0 1.8 1.5 3.1 59.5 0.3 10

0.7 0.6 2.1 3.9 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Autos&vehicles 10 10 10 10 10 10 10 10 10 10 10 10 10 10

1.1 3.1 11 4.9

3.3 3.9 3.1 8.6 Number of shares/views #views Sports

2.7 3.7 4.9 9.4

4.9 3.5 1.8 6.8

Life (a) CDF of #shares/#views. (b) Relationship between #views and

2.5 3.7 5.2 6.9

12 18 3.8 10.6 #shares.

Entertainment

4.1 3.8 3.3 10.1

0 0 0 0

Advertisement

1.7 3.9 7.9 2.5 Fig. 5. Videos in Renren.

3.1 4.1 3.5 23.2

Arts

2.7 4.1 5.4 32.2

0 0 0 0

Maternity

4.6 4.3 3.4 2.6

9.6 7.3 1.9 3.3 News&politics popular ONSs in China (as famous as Facebook and Twitter

3.8 4.6 4.3 7.9

5.1 1.5 0.7 19.1 Education in U.S.). When a video is shared from an online video website

5.6 4.8 3.1 25.9

9.9 2.5 0.6 18.2

Animation to an OSN, the remaining behaviors related to this video is

20 5.1 0.9 22.4

9.7 6.1 1.6 44.7

TV series watching and sharing within the OSN. Hence, in the following,

11 5.2 1.7 45.3

1.5 12 21 6.2

Comedy if we don’t clarify, the sharing behavior means internal sharing

1.6 5.9 13 4.1

7.8 11 3.6 5.7

Original within an OSN, which is different from the external sharing

15 10 2.4 4.4

2.6 3.3 3.3 9.3

Games studied in previous.

4.9 10 7.4 7.8

20 16 2.1 5.7

Music 15 23 5.4 5.5 First, we provide a general picture about video views and

Fraction of videos Fraction of Avg. number of Average length shares in Renren. Fig. 5 shows the CCDF of the number of

per category external shares per external shares per video

(%) category (%) per video (x100) (minutes) shares and views for a video. We find that the majority of videos are not shared or watched by a lot of people, e.g., 80% Fig. 4. Categories of videos in Youku and Tudou. of videos are shared less than 29 times, and 80% of videos are viewed by Renren users less than 153 times, which is similar to the external video sharing in online video websites. They Next, the category of a video also affects the external both have a long heavy tail. When we further investigate the sharing a lot. To see this, we show category’s effect in Fig. 4. relationship between views and shares, we find it has a very We find that music and animation are the top-2 categories simple mathematic formula as shown in Fig. 5b. If denote the that have the highest percentages of videos for both Youku number of views by , and the number of shares by , then and Tudou. Most external links in Youku come from video v s the relationship between v and s is categories music, games, original, comedy, TV series, and the same top-5 categories in Tudou come from categories l =0.32v0.93, entertainment, music, comedy, original and Hot topic. The largest averages of the number of external shares per video which is very close to a linear relationship. are both comedy for Youku and Tudou. Next, we do a reverse engineering study to study where the Finally, we study what are these external links pointing videos come from. The answer to this question can give us a to, or what external sites are consuming these videos. Table rank of the popularity of online video websites. We summarize III states the most frequently external websites identified from the most popular video websites in Renren in Tab. IV. In fact, the external links in Youku and Tudou. Among the external using different popularity metrics will obtain different ranks. websites, Renren and Qzone consume the majority of the We find that videos from Youku have the largest amount, views videos, about 96% of the Youku videos and 75% of the Tudou and shares, e.g., 57% of the videos in Renren are from Youku. videos. Meanwhile, Renren and Qzone consume more videos These videos attract 71% of all the views and 66% of all the than microblogs such as Sina and Tencent Weibo. shares. However, when averaged to each video, videos from Ku6.com has the highest average views and shares, e.g., each B. Video Consuming in Renren and Sina Weibo video in Ku6.com can attract 1102 shares and 11215 views on average. Since videos from Ku6.com only account for 4.8%, Having understand the external video sharing in online the average figures indicate that these minority videos are video websites, we now move to study how videos are con- high quality. For videos from each video website, the average sumed in Renren and Sina Weibo, which are the two most number of views per share has much smaller variance than the 50 TABLE V TRAFFIC STATISTICS BY CONTENT TYPE. 40

30 Distribution of #bytes (%) Distribution of #flows (%) 20 Videos 32.0 18.8

10

#videos (%) Images 31.5 32.8

0 Text 11.2 9.5 Youku Tudou Sina Yinyuetai 56.com Ku6 Joy Applications 2.2 32.9

35 Others 22.8 6.1

Avg(# of sharings per video)

30

Avg(# of comments per video)

25 TABLE VI

20 TRAFFIC STATISTICS BY ONLINE VIDEO WEBSITES.

15

10 Website Distribution of Distribution of Download speed 5 #bytes (%) #flows (%) (kB/s) 0 Youku 39.5 23.0 50 Youku Tudou Sina Yinyuetai 56.com Ku6 Sohu Joy 56.com 5.4 6.5 25 Tudou 7.3 5.9 50 Fig. 6. Video sites in Sina Weibo Sina.video.com 16.3 1.5 57 Tv.sohu.com 8.8 1.5 121 Ku6.com 0.58 1.3 6.2 Joy.cn 0.04 0.16 6.0 average number of views per video, which indicates that the Yinyuetai.com 0.91 0.06 53 number of views is strongly related to the number of share Others 21.1 60.1 27 actions. Now we move to Sina Weibo, the largest microblog in (t−∆,t). When ∆=1(second), we find that 9.9% of the video China. In Sina Weibo, about 7.4% of the tweets contain a flows are OSN related video flows, which account for 25.1% video and the same number is 2.4% in the original tweets, of the traffic generated by all videos. When ∆=2(seconds), 11.2% in the retweets. About 85.8% of the video tweets are about 10.6% of the video flows are OSN related video flows, retweets. These figures illustrate that people are more reluctant which account for 26.0% of the traffic generated by all videos. to retweet a video tweet than uploading a video by himself. For each video website, the traffic and flow fractions are shown Weibo users can post tweets via various ways. Loginning in Table VII. For the shortage of space, in the following into an account from a browser and posting tweets is the most analysis we set ∆=1(second). common way. Except that, about 27.2% of the users post tweets Next, we study the traffic properties inside OSNs. The via mobile devices. And the top 10 most popular mobile OSs traffic distributions of OSN related video flows among OSNs or devices used by Sina Weibo users are iPhone(13.92%), are shown in Table VIII. We can see that videos flows are much Android(5.62%), Nokia(4.71%), iPad(1.72%), Java(1.30%), more popular in Renren and Qzone, than in Sina and Tencent Motorola(0.19%), HTC(0.15%), BlackBerry(0.14%), SonyEr- Weibo. This finding is consistent with the results shown in icsson(0.10%), and WindowsMobile(0.08%). There is a dif- Table III, which are obtained from the datasets collected from ference of posting behavior for video tweets and non-video online video websites Youku and Tudou. We further show the tweets. For non-video tweets, the proportion of postings via traffic distributions of OSN related video flows among the mobile devices is 27.4%. But for video tweets, the number is video websites for each OSN website in Table IX. We can only 0.94%. This indicates that mobile devices are still not see that the top-3 most popular video websites in each OSN convenient for posting video tweets. are Youku, Ku6.com, and Tudou, which is also consistent with Fig. 6 shows the top 8 most popular video sites in Sina the results shown in Table IV that are obtained from the dataset Weibo. Youku and Tudou are the top 2 video sites which collected from Renren. generate more than 70% of the videos in Sina Microblog. But the most retweeted and commented videos are came from V. CONCLUSIONS Sina itself, which takes account only 9.5% of the videos. This indicates that videos from Sina are more attractive than others. We comprehensively studied video sharing between online video websites and online social networks from three perspec- tive views: online video websites, online social networks, and C. A Peep from Campus HTTP Traffic Data

Finally, we study the interactions between online video TABLE VII websites and OSNs from a much lower level—the campus TRAFFICANDFLOWFRACTIONSOF OSN RELATED VIDEOS FOR EACH HTTP traffic. The Internet traffic information should also ONLINE VIDEO WEBSITE. reflect some properties of these interactions. To see this, we Traffic fraction (%) Flow fraction (%) first show the fraction of different traffic from a campus in Website Table V and Table VI. We can see that 32% of the traffic is ∆=1 ∆=2 ∆=1 ∆=2 Youku 19.8 20.8 10.6 11.3 made of video, and 39.5% of the video traffic comes from 56.com 27.6 28.1 15.6 16.7 Youku. The videos from Sohu have the highest download Tudou 53.9 55.0 17.1 18.4 Sina.video.com 21.7 22.9 32.2 33.4 speed. Tv.sohu.com 23.0 23.2 19.1 21.0 Ku6.com 61.3 61.5 17.3 19.4 For a video flow, denote its start time by t, then we say Joy.cn 40.9 41.0 21.1 21.8 that a flow is an OSN related video flow if the previous flow Yinyuetai.com 47.8 47.9 30.4 31.6 with the same source IP came from an OSN in time interval Others 25.9 26.2 7.3 7.9 TABLE IV ONLINEVIDEOWEBSITESOFVIDEOSSHAREDIN RENREN.

Website Fraction of views per Fraction of shares per Fraction of videos per Avg. shares per video Avg. views per video Avg. views per share website (%) website (%) website (%) per website per website per website Youku 71 66 57 522 3942 7.5 Ku6.com 12 16 4.8 1102 11215 10 Tudou 9.2 9.6 26 150 1262 8.4 56.com 2.3 2.6 6.3 151 1404 9.3 Sina.video.com 1.4 1.8 0.95 618 6497 11 Joy.cn 1.0 1.5 0.93 463 5607 12 Yinyuetai.com 0.95 1.0 1.2 337 2932 8.7 Tv.sohu.com 0.40 0.73 0.88 191 2854 15 Others 1.0 0.81 1.9 225 1485 6.6

TABLE IX TRAFFICDISTRIBUTIONSOF OSN RELATEDVIDEOFLOWSAMONGONLINEVIDEOWEBSITESFOREACH OSN WEBSITE.

Website Youku Ku6.com Tudou 56.com Sina.video.com Joy.cn Yinyuetai.com Tv.sohu.com Others Renren 43.9 1.9 19.2 5.4 9.5 0.1 0.4 1.4 18.2 Qzone.qq.com 17.4 1.1 15.8 6.7 8.9 0 5.0 13.0 32.0 Kaixin001.com 40.3 0 0 9.5 11.2 2.1 0 0.1 36.8 Dist. of #bytes (%) Douban.com 17.4 0.5 3.9 6.0 11.3 0 0 26.6 34.4 Sina weibo (weibo.com) 10.2 0.3 3.6 4.8 42.4 0 1.7 24.5 12.5 Tencent weibo (t.qq.com) 20.6 0.4 16.7 11.5 8.5 0 1.1 5.6 35.7 Renren 47.7 3.9 7.1 8.2 3.5 0.5 0.1 1.2 27.8 Qzone.qq.com 9.0 1.3 13.7 10.7 3.2 0.3 0.3 3.5 58.1 Kaixin001.com 22.5 0 8.0 3.1 5.0 3.1 0 1.7 56.6 Dist. of #flows (%) Douban.com 11.1 0.8 6.1 15.5 6.1 0 0 5.4 54.9 Sina weibo (weibo.com) 9.0 0.7 12.0 10.1 15.7 0.3 0.3 5.8 46.2 Tencent weibo (t.qq.com) 18.6 0.4 5.0 18.8 3.1 0.1 0 2.3 51.6

TABLE VIII TRAFFICDISTRIBUTIONSOF OSN RELATED VIDEO FLOWS AMONG OSNS. [3] M. Cha, H. Kwak, P. Rodriguez, Y. Y. Ahn, and S. Moon, “I tube, you tube, everybody tubes: analyzing the world’s largest user generated content video system,” in Proceedings of ACM SIGCOMM IMC 2007, Website Distribution of #bytes Distribution of October 2007, pp. 1–14. (%) #flows(%) Renren 54.7 38.5 [4] X. Cheng, C. Dale, and J. Liu, “Statistics and of Qzone.qq.com 24.6 39.9 videos,” in Proceedings of IEEE IWQoS 2008, June 2008, pp. 229–238. Kaixin001.com 0.4 0.9 [5] G. Chatzopoulou, C. Sheng, and M. Faloutsos, “A first step towards Douban.com 1.0 1.9 understanding popularity in youtube,” in Proceedings of IEEE Infocom Sina Weibo (weibo.com) 4.4 6.7 2010 Network Science Workshop, March 2010, pp. 1–6. Tencent Weibo (t.qq.com) 14.5 12.1 [6] P. Gill, M. Arlitt, Z. Li, and A. Mahanti, “Youtube traffic characteriza- tion: A view from the edge,” in Proceedings of ACM SIGCOMM IMC 2007, October 2007, pp. 15–28. network traffic of a campus network. The results of our study [7] M. Zink, K. Suh, Y. Gu, and J. Kurose, “Characteristics of youtube provide insights on the interplays between the two kinds of network traffic at a campus network-measurements, models, and im- websites, which are summarized bellow: plications,” Journal Computer Networks, vol. 53, no. 4, pp. 501–514, March 2009. a) Many factors can affect the external sharing probability [8] M. Saxena, U. Sharan, and S. Fahmy, “Analyzing video services in of videos in online video websites. The nature of a video, e.g., web 2.0: a global perspective,” in In NOSSDAV 2008: Proceedings of its category, is also an important factor. the 18th International Workshop on Network and Operating Systems Support for Digital Audio and Video, May 2008, pp. 39–44. b) The popularity of a video itself in online video websites [9] V. K. Adhikari, S. Jain, and Z. Zhang, “Youtube traffic dynamics and can greatly impact on its popularity in OSNs. Videos in Renren its interplay with a tier-1 isp: An isp perspective,” in Proceedings of ACM SIGCOMM IMC 2010, November 2010, pp. 431–443. and Qzone usually attract more viewers than in Sina and [10] R. Torres, A. Finamore, J. Kim, M. Mellia, M. Munafo, and S. Rao, Tencent Weibo, which indicates the different natures of the “Dissecting video server selection strategies in the youtube cdn,” in two kinds of OSNs. Proceedings of IEEE ICDCS 2011, June 2011, pp. XX–XX. [11] K. Lai and D. Wang, “A measurement study of external links of c) 10% of video flows are associated with online social youtube,” in Proceedings of IEEE GLOBECOM 2009, December 2009, networks, and they account for 25% of traffic generated by all pp. 1–6. videos. [12] ——, “Towards understanding the external links of video. sharing sites: Measurement and analysis,” in In NOSSDAV 2010: Proceedings of the 20th International Workshop on Network and Operating Systems REFERENCES Support for Digital Audio and Video, June 2010, pp. 69–74. [13] D. Achlioptas, D. Kempe, A. Clauset, and C. Moore, “On the bias [1] C. Labovitz, S. Iekel-Johnson, D. McPherson, J. Oberheide, and F. Jaha- of traceroute sampling: or, power-law degree distributions in regular nian, “Internet inter-domain traffic,” in Proceedings of ACM SIGCOMM graphs,” in In ACM STOC, May 2005, pp. 694–703. 2010, September 2010, pp. 75–86. [14] M. Kurant, A. Markopoulou, and P. Thiran, “On the bias of bfs (breadth [2] comScore, “comScore releases febru- first search),” in Proceedings of International Teletraffic Congress 2010, ary 2013 u.s. online video rankings,” September 2010. http://www.comscore.com/Insights/Press Releases/2013/3/comScore Releases February 2013 U.S. Online Video Rankings, March 14 2013.

Top View