Presents corrections to the paper, “Modeling, analysis, and implementation of universal acceleration platform across online video sharing sites,” (Xu, K. et al), IEEE Trans. Serv. Comput., vol. 11, no. 3, pp. 534–548, May/Jun. 2018.

User-generated video sharing service has attracted a vast number of users over the Internet. The most successful sites, such as YouTube and Youku, now enjoy millions of videos being watched every day. Yet, given limited network and server resources, the user experience of existing video sharing sites (VSSes) is still far from being satisfactory. To mitigate such a problem, peer-to-peer (P2P) based video accelerators have been widely suggested to enhance the video delivery on VSSes. In this paper, we find that the interference of multiple accelerators will lead to a severe bottleneck across the VSSes. Our model analysis shows that a universal video accelerator can naturally achieve better performance with lower deployment cost. Based on this observation, we further present the detailed design of Peer-to-Peer Video Accelerator (PPVA), a real-world system for universal and transparent P2P accelerating. Such a system has already attracted over 180 million users, with 48 million video transactions every day. We carefully examine the PPVA performance from extensive measurements. Our trace analysis indicates that it can significantly reduce server bandwidth cost and accelerate the video download speed by 80 percent.

The deep penetration of Online Social Networks (OSNs) has made them as major portals for video information sharing. Propagated through chains of friends, the coverage of OSN-shared videos can be much broader with stronger micro- and macro-dynamics. Given that the contents are still hosted by external Video Sharing Sites (VSSes), such distinct access patterns from OSN users have created significant new challenges to VSSes. In this paper, we present SNACS, a cost-effective social network-aware cloud assistance for video sharing. The SNACS module sits between VSSes and an OSN, and is managed by the OSN to improve its users' video access experience using both centralized cloud resources and edge servers. Given the strong dynamics of the access patterns, we are particularly interested in the content management and update strategies in the SNACS' implementation. Motivated by real world data traces, we show that conventional cache replacement can be quite inefficient in this context. We then develop optimal offline algorithms with minimized cache misses and replacements, which also motivate an online solution that makes effective use of the video sharing patterns in the OSN. Our design has been extensively evaluated and its superiority has been validated under diverse network and user configurations.

This paper presents Pop-Forecast, a systematic method for accurately forecasting the popularity of videos promoted through social networks. Pop-Forecast aims to optimize the forecasting accuracy and the timeliness with which forecasts are issued, by explicitly taking into account the dynamic propagation of videos in social networks. The forecasting is performed online and requires no training phase or a priori knowledge. We analytically bound the performance loss of Pop-Forecast as compared to that obtained by an omniscient oracle and prove that the bound is sublinear in the number of video arrivals, thereby guaranteeing its fast rate of convergence as well as its asymptotic convergence to the optimal performance. We validate the performance of Pop-Forecast through extensive experiments using real-world data traces collected from the videos shared in RenRen, one of the largest online social networks in China. These experiments show that our proposed method outperforms existing approaches for popularity prediction (which do not take into account the propagation in social network) by more than 30% in terms of prediction rewards.

Modern online social networking has drastically changed the information distribution landscape. Recently, video has become one of the most important types of objects spreading among social networking service users. The sheer and ever-increasing data volume, the broader coverage, and the longer access durations of video objects, however, present significantly more challenges than other types of objects. This article takes an initial step toward understanding the unique characteristics of video sharing propagation in social networks. Based on realworld data traces from a large-scale online social network, we examine the user behavior from diverse aspects and identify different types of users involved in video propagation. We closely investigate the temporal distribution during propagation as well as the typical propagation structures, revealing more details beyond stationary coverage. We further extend the conventional epidemic models to accommodate diverse types of users and their probabilistic viewing and sharing behaviors. The model, effectively capturing the essentials of the propagation process, serves as a valuable basis for such applications as workload synthesis, traffic prediction, and resource provision of video servers.

Popularity prediction, with both technological and economic importance, has been extensively studied for conventional video sharing sites (VSSes), where the videos are mainly found via searching, browsing, or related links. Recent statistics however suggest that online social network (OSN) users regularly share video contents from VSSes, which has contributed to a significant portion of the accesses; yet the popularity prediction in this new context remains largely unexplored. In this paper, we present an initial study on the popularity prediction of videos propagated in OSNs along friendship links.

The social networking services (SNS) have drastically changed the information distribution landscape and people's daily life. With the development in broadband accesses, video has become one of the most important types of objects spreading among social networking service users, yet presents more significant challenges than other types of objects, not only to the SNS management, but also to the network traffic engineering. In this paper, we take an important step towards understanding the characteristics of video sharing propagation in SNS, based on the real viewing event traces from a popular SNS in China. We further extend the epidemic models to accommodate the diversity of the propagation, and our model effectively captures the propagation process of video sharing in SNS.

The deep penetration of Online Social Networks (OSNs) have made them major portals for video content sharing. It is known that a significant portion of the accesses to video sharing sites are now coming from OSN users. Yet the unique features of video sharing over OSNs and their impact remain largely unknown. In this paper, we present a measurement study towards understanding the video requests from OSNs. We closely collaborated with a large-scale Facebook-like OSN to analyze its user access logs spanning over four months. Our measurement reveals a number of distinctive features on the popularity distribution of videos shared over the OSN. In particular, we observe that the OSN amplifies the skewness of video popularity so largely that about 2% most popular videos account for 90% of total views; the video requests distribution also exhibits perfect powerlaw feature; video popularity evolution shows more dynamics. All these noticeably differ from that of conventional videos, such as YouTube videos. To further understand the characteristics, we model the video viewing and sharing behaviors in OSNs, leading to the development of a practical emulator. It reveals the gap between the sharing rate and the viewing rate, and generates user requests that well capture the video popularity distribution and dynamics as observed in our empirical data.

Today's peer-to-peer (P2P) Video-on-Demand (VoD) systems are known to be highly scalable in a steady state. For the dynamic scenario, much effort has been spent on accommodating sharply increasing requests (known as flash crowd) with effective solutions being developed. The high popularity upon a flash crowd however does not necessarily last long, and indeed often drops very fast after the peak. Compared to growth, a decay is seemingly less challenging or even beneficial given the less user demands. While this is true in a conventional client/server system, we find that it is not the case for peer-to-peer. A quick decay can easily de-stabilize an established overlay, and the resultant smaller overlay is generally less effective for content sharing. The replication of data segments, which is critical during flash crowd, will not promptly respond to a fast and globalized population decay, either. Many of the replicas can become redundant and, even worse, their spaces cannot be utilized for an extended period. In this paper, we seek to understand the impact of such decays and the key influential factors. To this end, we develop a mathematical model to trace the evolution of peer upload and replication during population churns, specifically during decays. Our model captures peer behaviors with common data replication and scheduling strategies in state-of-the-art peer-to-peer VoD systems. It quantitatively reveals the root causes toward escalating server load during a population decay. The model also facilitates the design of a flexible server provision to serve highly time-varying demands.

Recent statistics suggest that online social network (OSN) users regularly share video contents from video sharing sites (VSSes), and a significant amount of views of VSSes are indeed from OSN users nowadays. By crawling and comparing the statistics of same videos shared in both RenRen (the largest Facebook-like OSN in China) and Youku (the largest Youtube-like VSS in China), we find that the huge and distinguished video requests from OSNs have substantially changed the workload of VSSes. In particular, OSNs amplify the skewness of video popularity so largely that about 0.31% most popular videos account for 80% of total views. Another interesting phenomenon is that many popular videos in VSSes may not receive many requests in OSNs. To further understand these findings, we track the propagation process of videos shared in RenRen since their introduction to this OSN, and analyze the effect of potential parameters to such process, including the number of initiators (users who bring the video to the OSN directly from a VSS), branching factor (the number of users who watch the friend's shared video), and share rate (the probability that the viewers of a video will further share this video). Beyond our expectation, none of these factors determine a video's popularity in an OSN. Instead, it shows great randomness for the number of a video's potential requests when it is shared to an OSN. By modifying the basic Galton-Watson stochastic branching process, we develop a simple yet effective model to simulate the video propagation process in an OSN. Simulation results show that it can well capture the randomness of a video's popularity and the skewed video popularity distribution.

Youtube-like video sharing sites (VSSes) have gained increasing popularity in recent years. Meanwhile, Facebook-like online social networks (OSNs), have seen their tremendous success in connecting people of common interests. These two new generation of networked services are now bridged in that many users of OSNs share video contents originating from VSSes with their friends, and it has been shown that a significant portion of views of VSSes are attributed to this sharing scheme of social networks. To understand how the video sharing behavior, which is largely based on social relationship, impacts users' viewing pattern, we have conducted a long-term measurement with RenRen and YouKu, the largest online social network and the largest video sharing site in China, respectively. We show that social friends are more likely to have common interests and their sharing behaviors provide guidance to enhance recommended video lists. In this paper, we take a first step toward learning OSN video sharing patterns for VSS video recommendation. An auto-encoder model is developed to learn the social similarity of different videos in terms of their sharing in OSN. We therefore propose a similarity-based strategy to enhance recommended video lists for VSSes. Evaluation results demonstrate that this strategy can remarkably improve the precision in VSSes, as compared to state-of-the-art strategies without social information.

Online social networks (OSNs) have become popular destinations for connecting friends and sharing information. Recent statistics suggest that OSN users regularly share contents from video sites, and a significant amount of requests of the video sites are indeed from them nowadays. These behaviors have substantially changed the workload of online video services. To better understand this paradigm shift, we conduct a long-term and extensive measurement of video sharing in RenRen, the largest Facebook-like OSN in China. In this paper, we focus on the video popularity distribution and evolution. In particular, we find that the video popularity distribution exhibits perfect power-law feature (while videos in YouTube exhibit a power-law waist with a long truncated tail). Moreover, we observe that the requests for the new published videos generally experience two or three days latency to reach the peak value, and then change dynamically with a series of unpredictable bursts (while in YouTube, videos reach the global peak immediately after introduction to the system, and then the accesses generally decrease overtime, except possibly on some special days). These differences can raise new challenges to content providers. For example, the video popularity is now hard to predict based on their historical requests. We further develop a simple yet effective model to simulate user requests process across videos in OSNs. Trace-based simulation shows that it can well capture the observed features.

Since user demand for a Video-on-demand (VoD) service varies with time in one-day period, provisioning self-owned servers for the peak load it must sustain a few hours per day leads to bandwidth underutilization at other times. Content clouds, e.g. Amazon Cloud Front and Azure CDN, let VoD providers pay by bytes for bandwidth resources, potentially leading to cost savings even if the unit rate to rent a machine from a cloud provider is higher than the rate to own one. In this paper, based on long-term traces from two large-scale VoD systems and temporal development model of content clouds, we tackle challenges, design and potential benefits in migrating VoD services into the hybrid cloud-assisted deployment, where the user requests are partly served by the self-owned servers and partly served by the cloud. Our measurements show that the popularity of the most popular videos decays so quickly, for example, by 11% after one hour that it poses large challenges on updating videos in the cloud. However, the trace-driven evaluations show that our proposed migration strategies (active, reactive and smart strategies), although simply based on the current information, can make the hybrid cloud-assisted VoD deployment save up to 30% bandwidth expense compared with the Clients/Server mode. They can also handle unpredicted the flash crowd traffic with little cost. It also shows that the cloud price and server bandwidth chosen play the most important roles in saving cost, while the cloud storage size and cloud content update strategy play the key roles in the user experience improvement.

To alleviate the server bandwidth cost of online video sharing services, p2p delivering has been suggested as an effective tool with success already seen in accelerating individual sites. The numerous video sharing sites existed however call for a universal solution that provides transparent p2p acceleration beyond ad hoc solutions. More importantly, only a universal platform can fully explore the aggregated video and client resources across sites, particular for identical videos replicated in diverse sites. To this end, we develop PPVA, a working platform for universal and transparent P2SP (peer to server and peer) accelerating. As of May 2011, it has attracted over 190 million distinct clients, with 78 million daily transactions. We will demonstrate the novel features, implementation, and also effectiveness of PPVA.

Recent years have witnessed an explosion of online video sharing as a new killer Internet application. Yet, given limited network and server resources, user experience with existing video sharing sites are far from being satisfactory. To alleviate the bottleneck, peer-to-peer delivering has been suggested as an effective tool with success already seen in accelerating individual sites. The numerous video sharing sites existed however call for a universal solution that provides transparent peer-to-peer acceleration beyond ad hoc solutions. More importantly, only a universal platform can fully explore the aggregated video and client resources across sites, particular for identical videos replicated in diverse sites. To this end, we develop PPVA, a working platform for universal and transparent peer-to-peer accelerating. PPVA was first released in May 2008 and has since been constantly updated. As of January 2010, it has attracted over 50 million distinct clients, with 48 million daily transactions. In this paper, we highlight the unique challenges in implementing such a platform, and discuss the PPVA solutions. We have also constantly monitored the service of PPVA since its deployment. The mass amount of traces collected enables us to thoroughly investigate its effectiveness and potential drawbacks, and provide valuable guidelines to its future development.