Faster Web through Client-Assisted CDN Server Selection

Internet delivery infrastructures are traditionally optimized for low-latency traffic, such as the Web traffic. However, in recent years we are witnessing a massive growth of throughput-oriented applications, such as video streaming. These applications introduce new tradeoffs and design choices for content delivery networks (CDNs). In this paper, we focus on understanding two key design choices: (1) What is the impact of the number of CDN's peering points and server locations on its aggregate throughput and operating costs? (2) How much can ISP-CDNs benefit from using path selection to maximize its aggregate throughput compared to other CDNs who only have control at the edge? Answering these questions is challenging because content distribution involves a complex ecosystem consisting of many parties (clients, CDNs, ISPs) and depends on various settings which differ across places and over time. We introduce a simple model to illustrate and quantify the essential tradeoffs in CDN designs. Using extensive analysis over a variety of network topologies (with varying numbers of CDN peering points and server locations), operating cost models, and client video streaming traces, we observe that: (1) Doubling the number of peering points roughly doubles the aggregate throughput over a wide range of values and network topologies. In contrast, optimal path selection improves the CDN aggregate throughput by less than 70\%, and in many cases by as little as a few percents. (2) Keeping the number of peering points constant, but reducing the number of location (data centers) at which the CDN is deployed can significantly reduce operating costs.

The Internet architecture, based on end-to-end connections, had difficulties to efficiently deliver the always increasing number of contents. Content Delivery Networks (CDN) have been deployed to improve the delivery. Recent research works propose a new networking architecture, much more adapted to the current Internet usage (end-users just care about the contents they want and not about the endpoints that provide them). The Content-Centric Networking (CCN) aims at replacing the IP paradigm. CCN has been investigated by the research community for few years, and some demonstrators have proved its feasibility. However, less works have addressed the CDN use-case so far. In this paper, we show that the current CCN design, which does not allow negative reply, is not suitable for interconnection with the CDN service. We then propose the use of two new tables in the CCN nodes that are interconnected to the CDN servers, in order to detect possible misses in the CDN servers and subsequently forward the requests for missing contents toward the original servers instead of the CDN surrogates. The evaluations we have performed highlight that the integration of both new tables does not incur an increase of the memory requirement and thus a cost in the node. The proposed solution is thus a viable solution to make CCN network work in close cooperation with CDN networks, each one keeping its specific business models and functions: transport for CCN, storage for CDN.

Today’s content delivery networks (CDNs) use large sets of globally distributed servers, advanced routing techniques, and dynamic server selection to provide users with low delay, reliable access to Web, and streaming content. These strategies allow CDNs to extend their reach and performance into large numbers of networks around the globe. However, nearly 20% of Internet users reside in China, where local regulations and network policies make it challenging for global CDNs to serve foreign content from servers in China and limit the effectiveness of their traditional strategies for low-latency content delivery. Recently, a number of China-specific CDNs have partnered with global CDNs to provide service in the country. However, little is known about how CDNs in China are implemented, nor their impact on performance. In this paper, we are the first to investigate the impact of China’s unique policy and networking environment on CDN implementations and deployments. We find that deployments inside China exhibit different client-mapping behavior than global ones, namely, through static server selection for clients given a customer site, and through region-specific clients partitioning and mapping rather than using live network information. We also show that ignoring these properties can significantly impact server selection quality. Our results can be useful for optimization of CDNs in China.

Broadband network technologies have improved the bandwidth of the edge of the Internet but its core is still a bottleneck for large file transfers. Content Delivery Networks (CDNs), built at the edge of the Internet, are able to reduce the workload of network backbones but their scalability and network reach is often limited, especially in case of QoS-bound delivery services. By using the emerging CDN internetworking, a CDN can dynamically exploit resources of other cooperating CDNs to face peak loads and temporary malfunctions without violating QoS levels negotiated with content providers. This paper proposes an architectural schema and an algorithm to minimize transfer time of large files taking into account the SLA constraints agreed between the CDN in charge of content delivery and the content provider, while respecting the maximum budget that the CDN can pay to peer CDNs to ensure its revenue. The experimental analysis confirms the benefits of applying the proposed algorithm in CDN internetworking to enforce SLA guarantees of CDN services.

Internet-scale services like YouTube are provisioned by large Content Delivery Networks (CDNs), which push content as close as possible to the end-users to improve their Quality of Experience (QoE) and to pursue their own optimization goals. Adopting space and time variant traffic delivery policies, CDNs serve users' requests from multiple servers/caches at different physical locations and different times. CDNs traffic distribution policies can have a relevant impact on the traffic routed through the Internet Service Provider (ISP), as well as unexpected negative effects on the end-user QoE. In the event of poor QoE due to faulty CDN server selection, a major problem for the ISP is to avoid being blamed by its customers. In this paper we show a real case study in which Google CDN server selection policies negatively impact the QoE of the customers of a major European ISP watching YouTube. We argue that it is extremely important for the ISP to rapidly and automatically detect such events to increase its visibility on the overall operation of the network, as well as to promptly answer possible customer complaints. We therefore present an Anomaly Detection (AD) system for detecting unexpected cache-selection changes in the traffic delivered by CDNs. The proposed algorithm improves over traditional AD approaches by analyzing the complete probability distribution of the monitored features, as well as by self-adapting its functioning to dynamic environments, providing better detection capabilities.

Over the last years, the content delivery network (CDN) market has been witnessing globally an increase in the development of cloud-based CDNs, as they constitute a viable and cost-effective alternative to traditional commercial CDNs. Towards that direction, Social-Aware Virtual Network Embedding (SAViNE) for wireless content delivery within the OpenLab project aims at establishing, assessing, and prototyping a novel framework for deploying a CDN over the wireless cloud. Specifically, the goal of SAViNE is to provide experimental validation of the efficiency of social network analysis (SNA) - inspired surrogate server placement strategies, devised for fostering content delivery. The proposed framework has been integrated with the OpenLab experimental facilities, providing a run-time environment for CDN deployment, operation, and performance evaluation over the wireless environment. The framework’s feasibility and scalability is experimentally validated. The study is complemented with repeatability evaluation of experimental measurements, related to the resulting CDN’s operational efficiency.

The presence of bottlenecks in the Internet reduces the performance of content delivery in typical client-server applications such as Web browsing and media streaming. Caching and server farm architectures have been a good solution to this problem since the World Wide Web became very popular, but now they are not enough to support new e-business solutions. Content delivery networks (CDNs) distribute content by placing it on several surrogate servers across the Internet and try to overcome these bottlenecks. In CDN scenarios a good request routing algorithm selects the best surrogate for a client's request; it makes its decision using metrics received by the network and the surrogate. Typical metrics are network latency, hop count, available Internet bandwidth, and surrogate load. We analyze the impact of different request routing algorithms on the delivery performance of CDNs. A new request routing algorithm, called WSE (worst surrogate exclusion), is proposed. Worst surrogate exclusion excludes those surrogates that provide the client with low performance and prevents surrogates overloading. Simulation results show that WSE can reduce users average response time and efficaciously exploit the CDN's capacity.

Content Distribution Networks (CDN) appeared a decade ago as a method for reducing latencies, improving performance experienced by Internet users, and limiting the effect of flash-crowds, so as balance load in servers. Content Distribution has evolved in different ways (e.g. cloud computing structures and video streaming distribution infrastructures). The solution proposed in early CDN was the location of several controlled caching servers close to clients, organized and managed by a central control system. Many companies deployed their own CDN infrastructure– and so demonstrating the resulting effectiveness. However, the business model of these networks has evolved from the distribution of static web objects to video streaming. Many aspects of deployment and implementation remain proprietary, evidencing the lack of a general CDN model, although the main design concepts are widely known. In this work, the authors represent the structure of a CDN and the performance of some of its parameters, using queuing theory, simplifying the redirection schema and studying the elements that could determine the improvement in performance. The main contribution of the work is a general expression for a CDN environment and the relationship between different variables like caching hit ratios, network latency, number of surrogates, and server capacity; this proves that the use of CDN outperform the typical client/server architecture.

Recently, large-volume contents distributed by a content delivery network (CDN) on the Internet increase the load of content delivery servers and networks, which may degrade the quality of service. To overcome this problem, some mirror servers providing the same content are located on a network, and a request is navigated to one of the mirror servers. It is important to locate the mirror servers on the appropriate place in a network, as it affects the performance of the CDN. In this paper, we address the server location problem, which determines the location of the mirror servers satisfying the following two constraints: the number of the paths to the servers in a link is small, and the number of the nodes whose nearest mirror server is the same is small. The former constraint corresponds to the alleviation of the network load, and the latter constraint corresponds to the alleviation of the server load. First, we prove that this new server location problem is NP-complete. Next, we present a heuristic algorithm and evaluate it by applying to some actual network topologies. The results show that the algorithm can determine a good server location.

CDN (Content Delivery Network) has become an important infrastructure of the Internet. However, building an anomaly detection system to monitor and guarantee CDN service quality is non-trivial. Current anomaly detection system usually suffers from undesirable performance in terms of high rate of false positive and false negative, which consequently impacts on its practical deployment. Identifying the root cause of a false detection is critical for diagnosing and improving the performance of anomaly detection. In this paper, we propose a novel feedback mechanism for prediction-based anomaly detection in CDN . Specifically, we introduce a carefully-designed metric named Fittingscore to diagnose whether the prediction model can fit the data well. Further, a threshold adjustment mechanism is proposed to dynamically adjust the thresholds of residual errors. Extensive experiments employing a three-month real CDN dataset collected from a top ISP-operated CDN in China show our proposed method can significantly improve the performance of anomaly detection.

Content delivery over Internet is optimized for low use perceived latency by designing specialized Content Distribution Network (CDN). Content servers are connected through a content switch the distributes client requests among them to achieve load balancing across the servers. The content switch is located one-hop away from the servers. Server load balancing is one factor in achieveing low user perceived latency and improving operational efficiency. Traffic engineering is another factor that is generally and integrated with serverload balancing for CDN optimization. In this thesis we propose a request routing algorithm for a CDN that was designed to integrate server selection and traffic engineering functions in the request routing system. The CDN employs MPLS in the network for traffic engineering. The proposed algorithm optimizes content delivery for user perceived latency by achieving server load balancing and network traffic loan manaement among alternative paths. It also improves operational efficiency of the CDN by eliminating bottleneck paths and increasing utilization of underutilized servers and paths.

Content delivery networks (CDNs) provide an efficient support for serving http and streaming media content white minimizing the network impact of content delivery as well as overcoming the server overload problem. For serving the large documents and media files, there is an additional problem of the original content distribution across the CDN edge servers. We propose an algorithm, called ALM-fastreplica, for optimizing replication of large files across the edge servers in CDNs. The original file is partitioned into k subfiles, and each subfile is replicated via a correspondingly constructed multicast tree. Nodes from the different multicast trees use additional cross-nodes connections to exchange their corresponding subfiles such that each node eventually receives an entire file. This new replication method significantly reduces file replication time, up to 5-15 times compared to the traditional unicast (or point-to-point) schema. Since a single node failure in the multicast tree during the file distribution may impact the file delivery to a significant number of nodes, it is important to design an algorithm which is able to deal with node failures. We augment ALM-FastReplica with an efficient reliability mechanism, that can deal with node failures by making local repair decisions within a particular replication group of nodes. Under the proposed algorithm, the load of the failed node is shared among the nodes of the corresponding replication group, making the performance degradation gradual.

We analyze competition between Internet Service Providers (ISPs) where consumers demand heterogeneous content within two Quality-of-Service (QoS) regimes, Net Neutrality and Paid Prioritization, and show that paid prioritization increases the static efficiency compared to a neutral network. We also consider paid prioritization intermediated by Content Delivery Networks (CDNs). While the use of CDNs is welfare neutral, it results in higher consumer prices for internet access. Regarding incentives to invest in network capacity we show that discriminatory regimes lead to higher incentives than the neutral regime as long as capacity is scarce, while investment is highest in the presence of CDNs.

This paper describes issues in current multicast content delivery network (CDN) architecture and presents a framework for multicast CDN to satisfy the requirements for content delivery service (Hayashi et al., 2005). In our experience, requirements for CDN (Hayashi et al., 2005) are not perfectly covered by possible current methods using current group management protocols such as IGMP (B. Cain et al., 2002) and MLD (R. Vida et al., 2004). Therefore, we propose a new framework for multicast CDN to support authentication authorization & accounting (AAA) and QoS control

With the increasing popularity of the Internet video streaming services (e.g. YouTube and Netflix), content delivery networks (CDNs) are heavily used to stream video contents to users, and consume more and more power and bandwidths in recent years. In this paper, we investigate the problem of saving a video streaming CDN’s operating expense, including both its energy cost and the traffic cost. From our measurement study on the CDN infrastructure of Youku, which is the largest video site in China, we find that there exists an inherent conflict between improving a video streaming CDN’s energy efficiency for power saving, and maintaining the CDN’s ISP-friendly server selection policy. To address this problem, we propose a cost-aware capacity provisioning algorithm, which dynamically plans the service capacities of a CDN’s server clusters in numerous ISPs, and optimizes its overall operating cost regarding both the energy consumptions and the cross-ISP traffics. By using the workload derived from real-world measurement and applying actual power and bandwidth price parameters, we show with experiments that our approach can significantly reduce a video streaming CDN’s overall operating cost, and avoid frequent server switches effectively. To our best knowledge, this work is the first one that identifies and resolves the inherent conflict between a CDN’s energy efficiency and its ISP-friendly policy.