Cloud computing enables the dynamic provisioning of server resources. To exploit this opportunity, a policy is needed for dynamically allocating (and deallocating) servers in response to the current load conditions. In this paper we describe several simple policies for dynamic server allocation and develop analytic models for their analysis. We also design semi-Markov decision models that enable determination of the performance achieved with optimal policies, allowing us to quantify the performance gap between simple, easily implemented policies, and optimal policies. Finally, we apply our models to study the potential performance benefits of state-dependent routing in multi-site systems when using dynamic server allocation at each site. Insights from our results are valuable to service providers wanting to balance cloud service costs and delays.

Content-delivery applications can achieve scalability and reduce wide-area network traffic using geographically distributed caches. However, each deployed cache has an associated cost, and under time-varying request rates (e.g., a daily cycle) there may be long periods when the request rate from the local region is not high enough to justify this cost. Cloud computing offers a solution to problems of this kind, by supporting dynamic allocation and release of resources. In this article, we analyze the potential benefits from dynamically instantiating caches using resources from cloud service providers. We develop novel analytic caching models that accommodate time-varying request rates, transient behavior as a cache fills following instantiation, and selective cache insertion policies. Within the context of a simple cost model, we then develop bounds and compare policies with optimized parameter selections to obtain insights into key cost/performance tradeoffs. We find that dynamic cache instantiation can provide substantial cost reductions, that potential reductions strongly dependent on the object popularity skew, and that selective cache insertion can be even more beneficial in this context than with conventional edge caches. Finally, our contributions also include accurate and easy-to-compute approximations that are shown applicable to LRU caches under time-varying workloads.

The demand and usage of 360° video services are expected to increase. However, despite these services being highly bandwidth intensive, not much is known about the potential value that basic bandwidth saving techniques such as server or edge-network on-demand caching (e.g., in a CDN) could have when used for delivery of such services. This problem is both important and complicated as client-side solutions have been developed that split the full 360° view into multiple tiles, and adapt the quality of the downloaded tiles based on the user’s expected viewing direction and bandwidth conditions. This article presents new trace-based analysis methods that incorporate users’ viewports (the area of the full 360° view the user actually sees), a first characterization of the cross-user similarities of the users’ viewports, and a trace-based analysis of the potential bandwidth savings that caching-based techniques may offer under different conditions. Our analysis takes into account differences in the time granularity over which viewport overlaps can be beneficial for resource saving techniques, compares and contrasts differences between video categories, and accounts for uncertainties in the network conditions and the prediction of the future viewing direction when prefetching. The results provide substantial insight into the conditions under which overlap can be considerable and caching effective, and inform the design of new caching system policies tailored for 360° video.

The demand and usage of 360-degree video services are expected to increase. However, despite these services being highly bandwidth intensive, not much is known about the potential value that basic bandwidth saving techniques such as server or edge-network on-demand caching (e.g., in a CDN) could have when used for delivery of such services. This problem is both important and complicated as client-side solutions have been developed that split the full 360-degree view into multiple tiles, and adapt the quality of the downloaded tiles based on the user's expected viewing direction and bandwidth conditions. To better understand the potential bandwidth savings that caching-based techniques may offer for this context, this paper presents the first characterization of the similarities in the viewing directions of users watching the same 360-degree video, the overlap in viewports of these users (the area of the full 360-degree view they actually see), and the potential cache hit rates for different video categories and network conditions. The results provide substantial insight into the conditions under which overlap can be considerable and caching effective, and can inform the design of new caching system policies tailored for 360-degree video.

With 360° video, only a limited fraction of the full view is displayed at each point in time. This has prompted the design of streaming delivery techniques that allow alternative playback qualities to be delivered for each candidate viewing direction. However, while prefetching based on the user's expected viewing direction is best done close to playback deadlines, large buffers are needed to protect against shortfalls in future available bandwidth. This results in conflicting goals and an important prefetch aggressiveness tradeoff problem regarding how far ahead in time from the current play-point prefetching should be done. This paper presents the first characterization of this tradeoff. The main contributions include an empirical characterization of head movement behavior based on data from viewing sessions of four different categories of 360° video, an optimization-based comparison of the prefetch aggressiveness tradeoffs seen for these video categories, and a data-driven discussion of further optimizations, which include a novel system design that allows both tradeoff objectives to be targeted simultaneously. By qualitatively and quantitatively analyzing the above tradeoffs, we provide insights into how to best design tomorrow's delivery systems for 360° videos, allowing content providers to reduce bandwidth costs and improve users' playback experiences.

Video streaming applications generate a large fraction of Internet traffic. Much of this content is delivered over HTTP using standard web servers. Unlike other types of web workloads, HTTP video streaming workloads are typically disk bound, and therefore an important problem is that of optimizing disk access.

In contrast to traditional video, multi-view video streaming allows viewers to interactively switch among multiple perspectives provided by different cameras. One approach to achieve such a service is to encode the video from all of the cameras into a single stream, but this has the disadvantage that only a portion of the received video data will be used, namely that required for the selected view at each point in time. In this paper, we introduce the concept of a “multi-video stream bundle” that consists of multiple parallel video streams that are synchronized in time, each providing the video from a different camera capturing the same event or movie. For delivery we leverage the adaptive features and time-based chunking of HTTP-based adaptive streaming, but now employing adaptation in both content and rate. Users are able to change their viewpoint on-demand and the client player adapts the rate at which data are retrieved from each stream based on the user's current view, the probabilities of switching to other views, and the user's current bandwidth conditions. A crucial component of such a system is the prefetching policy. For this we present an optimization model as well as a simpler heuristic that can balance the playback quality and the probability of playback interruptions. After analytically and numerically characterizing the optimal solution, we present a prototype implementation and sample results. Our prefetching and buffer management solution is shown to provide close to seamless playback switching when there is sufficient bandwidth to prefetch the parallel streams.

The ephemeral content popularity seen with many content delivery applications can make indiscriminate on-demand caching in edge networks highly inefficient, since many of the content items that are added to the cache will not be requested again from that network. In this paper, we address the problem of designing and evaluating more selective edge-network caching policies. The need for such policies is demonstrated through an analysis of a dataset recording YouTube video requests from users on an edge network over a 20-month period. We then develop a novel workload modelling approach for such applications and apply it to study the performance of alternative edge caching policies, including indiscriminate caching and cache on kth request for different k. The latter policies are found able to greatly reduce the fraction of the requested items that are inserted into the cache, at the cost of only modest increases in cache miss rate. Finally, we quantify and explore the potential room for improvement from use of other possible predictors of further requests. We find that although room for substantial improvement exists when comparing performance to that of a perfect “oracle” policy, such improvements are unlikely to be achievable in practice.

This paper considers the problem of providing users playing one streaming video the option of instantaneous and seamless playback of alternative videos. Recommendation systems can easily provide a list of alternative videos, but there is little research on how to best eliminate the startup time for these alternative videos. The problem is motivated by services that want to retain increasingly impatient users, who frequently watch the beginning of multiple videos, before viewing a video to the end. We present the design, implementation, and evaluation of an HTTP-based Adaptive Streaming (HAS) solution that provides careful prefetching and buffer management. We also present the design and evaluation of three fundamental policy classes that provide different tradeoffs between how aggressively new alternative videos are prefetched versus the importance of ensuring high playback quality. We show that our solution allows us to reduce the startup times of alternative videos by an order of magnitude and effectively adapt the quality such as to ensure the highest possible playback quality of the video being viewed. By improving the channel utilization we also address the discrimination problem that HAS clients often suffer from, allowing us to in some cases simultaneously improve the playback quality of the video being viewed and provide the value-added service of allowing instantaneous playback of the prefetched alternative videos.

Interactive branched video that allows users to select their own paths through the video, provides creative content designers with great personalization opportunities; however, such video also introduces significant new challenges for the system developer. For example, without careful prefetching and buffer management, the use of multiple alternative playback paths can easily result in playback interruptions. In this paper, we present a full implementation of an interactive branched video player using HTTP-based Adaptive Streaming (HAS) that provides seamless playback even when the users defer their branch path choices to the last possible moment. Our design includes optimized prefetching policies that we derive under a simple optimization framework, effective buffer management of prefetched data, and the use of parallel TCP connections to achieve efficient buffer workahead. Through performance evaluation under a wide range of scenarios, we show that our optimized policies can effectively prefetch data of carefully selected qualities along multiple alternative paths such as to ensure seamless playback, offering users a pleasant viewing experience without playback interruptions.

Past work has shown that disk prefetching can be an effective technique for improving the performance of disk bound workloads. However, the performance gains are highly dependent on selecting a prefetch size that is appropriate for a specific system and workload. Using a prefetch size that is too small can lead to poor overall disk throughput, whereas prefetch sizes that are too large can lead to data being evicted before it can be used by a subsequent request.

This paper presents new results on characterization and modeling of user-generated video popularity evolution, based on a recent complementary data collection for videos that were previously the subject of an eight month data collection campaign during 2008/09. In particular, during 2011, we collected two contiguous months of weekly view counts for videos in two separate 2008/09 datasets, namely the ``recently-uploaded'' and the ``keyword-search'' datasets. These datasets contain statistics for videos that were uploaded within 7 days of the start of data collection in 2008 and videos that were discovered using a keyword search algorithm in 2008, respectively. Our analysis shows that the average weekly view count for the recently-uploaded videos had not decreased by the time of the second measurement period, in comparison to the middle and later portions of the first measurement period. The new data is used to evaluate the accuracy of a previously proposed model for synthetic view count generation for time periods that are substantially longer than previously considered. We find that the model yielded distributions of total (lifetime) video view counts that match the empirical distributions, however, significant differences between the model and empirical data were observed with respect to other metrics. These differences appear to arise because of particular popularity characteristics that change over time rather than being week-invariant as assumed in the model.

HTTP-based Adaptive Streaming (HAS) has become a widely-used video delivery technology. Use of HTTP enables relatively easy firewall/NAT traversal and content caching. While caching is an important aspect of HAS, there is not much public research on the performance impact proxies and their policies have on HAS. In this paper we build an experimental framework using open source Squid proxies and the most recent Open Source Media Framework (OSMF). A range of content-aware policies can be implemented in the proxies and tested, while the player software can be instrumented to measure performance as seen at the client. Using this framework, the paper makes three main contributions. First, we present a scenario-based performance evaluation of the latest version of the OSMF player. Second, we quantify the benefits using different proxy-assisted solutions, including basic best effort policies and more advanced content quality aware prefetching policies. Finally, we present and evaluate a cooperative framework in which clients and proxies share information to improve performance. In general, the bottleneck location and network conditions play central roles in which policy choices are most advantageous, as they significantly impact the relative performance differences between policy classes. We conclude that careful design and policy selection is important when trying to enhance HAS performance using proxy assistance.

Video dissemination through sites such as YouTube can have widespread impacts on opinions, thoughts, and cultures. Not all videos will reach the same popularity and have the same impact. Popularity differences arise not only because of differences in video content, but also because of other "content-agnostic" factors. The latter factors are of considerable interest but it has been difficult to accurately study them. For example, videos uploaded by users with large social networks may tend to be more popular because they tend to have more interesting content, not because social network size has a substantial direct impact on popularity. In this paper, we develop and apply a methodology that is able to accurately assess, both qualitatively and quantitatively, the impacts of various content-agnostic factors on video popularity. When controlling for video content, we observe a strong linear "rich-get-richer" behavior, with the total number of previous views as the most important factor except for very young videos. The second most important factor is found to be video age. We analyze a number of phenomena that may contribute to rich-get-richer, including the first-mover advantage, and search bias towards popular videos. For young videos we find that factors other than the total number of previous views, such as uploader characteristics and number of keywords, become relatively more important. Our findings also confirm that inaccurate conclusions can be reached when not controlling for content.

Wireless sensor networks are often deployed to monitor scalar fields. Commonly, these networks employ a query-based architecture, where specific data are requested by the sink node, and the network responds. Contours or isolines are compact representations of the state of a scalar field, and are useful abstractions of scalar field behavior. Previously researchers have examined algorithms and policies for detecting contours in query-based sensor networks, but have only addressed individual components of the process. In this paper we present a set of algorithms for contour detection in WSNs, and associated oracle functions to provide an unbiased optimality comparison. We show that our algorithms, and in particular our response and aggregation policy provide near-optimal performance in simulated networks measuring physically realistic scalar fields.

One highly-scalable approach to content delivery is to harness the upload bandwidth of the clients. Peer-assisted content delivery systems have been shown to effectively offload the servers of popular files, as the request rates of popular content enable the formation of self-sustaining torrents, where the entire content of the file is available among the peers themselves. However, for less popular files, these systems are less helpful in offloading servers. With a long tail of mildly popular content, with a high aggregate demand, a large fraction of the file requests must still be handled by servers. In this paper, we present the design, implementation, and evaluation of a dynamic file bundling system, where peers are requested to download content which they may not otherwise download in order to “inflate” the popularity of less popular files. Our system introduces the idea of a super bundle, which consists of a large catalogue of files. From this catalogue, smaller bundles, consisting of a small set of files, can dynamically be assigned to individual users. The system can dynamically adjust the number of downloaders of each file and thus enables the popularity inflation to be optimized according to current file popularities and the desired tradeoff between download times and server resource usage. The system is evaluated on PlanetLab.

Large amounts of Internet streaming video traffic are being delivered using HTTP to leverage the existing web infrastructure. A fundamental issue in HTTP streaming concerns the granularity of video objects used throughout the HTTP ecosystem (including clients, proxy caches, CDN nodes, and servers). A video may be divided into many files (called chunks), each containing only a few seconds of video at one extreme, or stored in a single unchunked file at the other.

With the proliferation of cloud services, cloud-based systems can become a cost-effective means of on-demand content delivery. In order to make best use of the available cloud bandwidth and storage resources, content distributors need to have a good understanding of the tradeoffs between various system design choices. In this work we consider a peer-assisted content delivery system that aims to provide guaranteed average download rate to its customers. We show that bandwidth demand peaks for contents with moderate popularity, and identify these contents as candidates for cloud-based service. We then consider dynamic content bundling (inflation) and cross-swarm seeding, which were recently proposed to improve download performance, and evaluate their impact on the optimal choice of cloud service use. We find that much of the benefits from peer seeding can be achieved with careful torrent inflation, and that hybrid policies that combine bundling and peer seeding often reduce the delivery costs by 20% relative to only using seeding. Furthermore, all these peer-assisted policies reduce the number of files that would need to be pushed to the cloud. Finally, we show that careful system design is needed if locality is an important criterion when choosing cloud-based service provisioning.

Recent increases in live and on-demand video streaming have dramatically changed the Internet landscape. In North America, Netflix alone accounts for 28% of all and 33% of peak downstream Internet traffic on fixed access links, with further rapid growth expected [26]. This increase in streaming traffic coincides with the steady adoption of HTTP for use in video streaming. Many streaming video providers, such as Apple, Adobe, Akamai, Netflix and Microsoft, now use HTTP to stream content [5]. Therefore, it is critical that we understand the impact of this emerging workload on web servers. Unlike other web content, a recent study [13] of streaming video shows that even small infrequent latency spikes, manifested as buffering related pauses, can result in shorter viewing times especially during live broadcasts. Unfortunately, no appropriate benchmarks exist to evaluate web servers under HTTP video streaming workloads.

Peer-assisted content delivery systems can provide scalable download service for popular files. For mildly popular content, however, these systems are less helpful in offloading servers as the request rate for less popular files may not enable formation of self-sustaining torrents (where the entire content of the file is available among the peers themselves). As there typically is a long tail of mildly popular content, with a high aggregate demand, a large fraction of the file requests must still be handled by servers, and is not off-loadable to peers. Bundling approaches have been proposed where peers are requested to download content which they may not otherwise be interested in order to ``inflate'' the popularity of less popular files. We present the design and implementation of a dynamic bundling system, in which a large number of files may be bundled to form a super bundle. From this super bundle, smaller individual bundles, consisting of a small set of files, can dynamically be assigned to individual users. Our system has the capability to dynamically adjust the number of downloaders of each file, thus allowing popularity inflation to be optimized according to current file popularities.

Video sharing services that allow ordinary Web users to upload video clips of their choice and watch video clips uploaded by others have recently become very popular. This article identifies invariants in video sharing workloads, through comparison of the workload characteristics of four popular video sharing services. Our traces contain metadata on approximately 1.8 million videos which together have been viewed approximately 6 billion times. Using these traces, we study the similarities and differences in use of several Web 2.0 features such as ratings, comments, favorites, and propensity of uploading content. In general, we find that active contribution, such as video uploading and rating of videos, is much less prevalent than passive use. While uploaders in general are skewed with respect to the number of videos they upload, the fraction of multi-time uploaders is found to differ by a factor of two between two of the sites. The distributions of lifetime measures of video popularity are found to have heavy-tailed forms that are similar across the four sites. Finally, we consider implications for system design of the identified invariants. To gain further insight into caching in video sharing systems, and the relevance to caching of lifetime popularity measures, we gathered an additional dataset tracking views to a set of approximately 1.3 million videos from one of the services, over a twelve-week period. We find that lifetime popularity measures have some relevance for large cache (hot set) sizes (i.e., a hot set defined according to one of these measures is indeed relatively “hot”), but that this relevance substantially decreases as cache size decreases, owing to churn in video popularity.

Sensor networks are often deployed more densely than would be minimally required. In such cases, node scheduling protocols can be used to determine which nodes are active, and which nodes sleep so as to conserve energy and prolong network lifetime. A drawback of node scheduling approaches, however, is delay due to node or communication failure(s), and subsequent wake-up of replacement node(s), during which monitoring coverage of some sub-region may be lost. This paper proposes an alternative approach for use in contexts in which the objective is to periodically collect sensing data that completely covers a region of interest. In the proposed approach, nodes dynamically determine during each round of data collection whether they should transmit their data, or whether their area is covered by neighbouring nodes that have already transmitted. Both unicast and broadcast-based data collection protocols are designed, and their performance compared using simulation to that of data collection protocols relying on node scheduling. Our results suggest that the coverage-preserving broadcast-based protocol can greatly improve reliability at the potential cost of increased traffic volume owing to non-minimal selection of transmitting nodes.

With a majority of Internet traffic being predicted to be caused by content delivery, it is clear that content delivery applications will consume much of the resources on the Internet. This paper considers the problem of cost-efficient content delivery, in which the application incurs both a network delivery cost (e.g., from cross ISP traffic or, more generally, operation/energy costs at Internet routers) and costs at the servers (e.g., due to cost of ownership, energy, or disk bandwidth). While the cost objective and the absolute cost tradeoff may be different from case to case, we argue that an architecture with distributed servers, each using digital fountain delivery, may be an attractive candidate architecture when considering the total content delivery cost. Within the context of a simple system model, we then determine optimal server selection policies for such an architecture, and derive analytic expressions for their associated delivery costs. A readily-implementable heuristic policy is proposed that is found to achieve within 10% of the minimal cost. Finally, we show how our results for content download can also be applied to streaming video delivery.

Video on demand, particularly with user-generated content, is emerging as one of the most bandwidth-intensive applications on the Internet. Owing to content control and other issues, some video-on-demand systems attempt to prevent downloading and peer-to-peer content delivery. Instead, such systems rely on server replication, such as via third-party content distribution networks, to support video streaming (or pseudostreaming) to their clients. A major issue with such systems is the cost of the required server resources. By synchronizing the video streams for clients that make closely spaced requests for the same video from the same server, server costs (such as for retrieval of the video data from disk) can be amortized over multiple requests. A fundamental trade-off then arises, however, with respect to server selection. Network delivery cost is minimized by selecting the nearest server, while server cost is minimized by directing closely spaced requests for the same video to a common server. This article compares classes of server selection policies within the context of a simple system model. We conclude that: (i) server selection using dynamic system state information (rather than only proximities and average loads) can yield large improvements in performance, (ii) deferring server selection for a request as late as possible (i.e., until just before streaming is to begin) can yield additional large improvements, and (iii) within the class of policies using dynamic state information and deferred selection, policies using only “local” (rather than global) request information are able to achieve most of the potential performance gains.

No abstract available.

Conventional video consists of a single sequence of video frames. During a client's playback period, frames are viewed sequentially from some specified starting point. The fixed frame ordering of conventional video enables efficient scheduled broadcast delivery, as well as efficient near on-demand delivery to large numbers of concurrent clients through use of periodic broadcast protocols in which the video file is segmented and transmitted on multiple channels. This paper considers the problem of devising scalable protocols for near on-demand delivery of "nonlinear" media files whose content may have a tree or graph, rather than linear, structure. Such media allows personalization of the media playback according to individual client preferences. We formulate a mathematical model for determination of the optimal periodic broadcast protocol for nonlinear media with piecewise-linear structures. Our objective function allows differing weights to be placed on the startup delays required for differing paths through the media. Studying a number of simple nonlinear structures we provide insight into the characteristics of the optimal solution. For cases in which the cost of solving the optimization model is prohibitive, we propose and evaluate an efficient approximation algorithm.

Periodic broadcast protocols enable efficient streaming of highly popular media files to large numbers of concurrent clients. Most previous periodic broadcast protocols, however, assume that all clients can receive at the same rate, and also assume that reception bandwidth is not time-varying. In this article, we first develop a new periodic broadcast protocol, Optimized Heterogeneous Periodic Broadcast (OHPB), that can be optimized for a given population of clients with heterogeneous reception bandwidths and quality-of-service requirements. The OHPB protocol utilizes an optimized segment size progression determined by solving a linear optimization model that takes as input the client population characteristics and an objective function such as mean client startup delay. We then develop a generalization of the OHPB linear optimization model that allows optimal server bandwidth allocation among multiple concurrent OHPB broadcasts, wherein each media file and its clients may have different characteristics. Finally, we propose complementary client protocols employing work-ahead buffering of data during playback, so as to enable more uniform playback quality when the reception bandwidth is time-varying.

Periodic broadcast protocols enable the efficient streaming of highly popular media files to large numbers of concurrent clients. Most previous periodic broadcast protocols, however, assume that all clients can receive at the same rate, and also assume that available bandwidth is not time-varying. In this paper, we first develop a new periodic broadcast protocol, Optimized Heterogeneous Periodic Broadcast (OHPB), that can be optimized for a given population of clients with heterogeneous reception bandwidths and quality-of-service requirements. The OHPB protocol utilizes an optimized segment size progression determined by solving a linear optimization model that takes as input the client population characteristics and an objective function such as mean client startup delay. We then propose complementary client protocols employing work-ahead buffering of data during playback, so as to enable more uniform playback quality when the available bandwidth is time-varying.

A conventional video file contains a single temporally-ordered sequence of video frames. Clients requesting on-demand streaming of such a file receive (all or intervals of) the same content. For popular tiles that receive many requests during a file playback time, scalable streaming protocols based on multicast or broadcast have been devised. Such protocols require server and network bandwidth that grow much slower than linearly with the file request rate. This paper considers "nonlinear" video content in which there are parallel sequences of frames. Clients dynamically select which branch of the video they wish to follow, sufficiently ahead of each branch point so as to allow the video to he delivered without jitter. An example might be "choose-your-own-ending" movies. With traditional scalable delivery architectures such as movie theaters or TV broadcasting, such personalization of the delivered video content is very difficult or impossible. It becomes feasible, in principle at least, when the video is streamed to individual clients over a network. This paper analyzes the minimal server bandwidth requirements, and proposes and evaluates practical scalable delivery protocols, for on-demand streaming of nonlinear media.

Previous scalable protocols for downloading large, popular files from a single server include batching and cyclic multicast. With batching, clients wait to begin receiving a requested file until the beginning of its next multicast transmission, which collectively serves all of the waiting clients that have accumulated up to that point. With cyclic multicast, the file data is cyclically transmitted on a multicast channel. Clients can begin listening to the channel at an arbitrary point in time, and continue listening until all of the file data has been received. This paper first develops lower bounds on the average and maximum client delay for completely downloading a file, as functions of the average server bandwidth used to serve requests for that file, for systems with homogeneous clients. The results show that neither cyclic multicast nor batching consistently yields performance close to optimal. New hybrid download protocols are proposed that achieve within 15 % of the optimal maximum delay and 20 % of the optimal average delay in homogeneous systems. For heterogeneous systems in which clients have widely-varying achievable reception rates, an additional design question concerns the use of high rate transmissions, which can decrease delay for clients that can receive at such rates, in addition to low rate transmissions that can be received by all clients. A new scalable download protocol for such systems is proposed, and its performance is compared to that of alternative protocols as well as to new lower bounds on maximum client delay. The new protocol achieves within 25 % of the optimal maximum client delay in all scenarios considered.

Recent scalable multicast streaming protocols for on-demand delivery of media content offer the promise of greatly reduced server and network bandwidth. However, a key unresolved issue is how to design scalable content distribution systems that place replica servers closer to various client populations and route client requests and response streams so as to minimize the total server and network delivery cost. This issue is significantly more complex than the design of distribution systems for traditional Web files or unicast on-demand streaming, for two reasons. First, closest server and shortest path routing does not minimize network bandwidth usage; instead, the optimal routing of client requests and server multicasts is complex and interdependent. Second, the server bandwidth usage increases with the number of replicas. Nevertheless, this paper shows that the complex replica placement and routing optimization problem, in its essential form, can be expressed fairly simply, and can be solved for example client populations and realistic network topologies. The solutions show that the optimal scalable system can differ significantly from the optimal system for conventional delivery. Furthermore, simple canonical networks are analyzed to develop insights into effective heuristics for near-optimal placement and routing. The proposed new heuristics can be used for designing large and heterogeneous systems that are of practical interest. For a number of example networks, the best heuristics produce systems with total delivery cost that is within 16% of optimality.

Previous scalable on-demand streaming protocols do not allow clients to recover from packet loss. This paper develops new protocols that: (1) have a tunably short latency for the client to begin playing the media; (2) allow heterogeneous clients to recover lost packets without jitter as long as each client's cumulative loss rate is within a tunable threshold; and (3) assume a tunable upper bound on the transmission rate to each client that can be as small as a fraction (e.g., 25%) greater than the media play rate. Models are developed to compute the minimum required server bandwidth for a given loss rate and playback latency. The results of the models are used to develop the new protocols and assess their performance. The new protocols, Reliable Periodic Broadcast and Reliable Bandwidth Skimming, are simple to implement and achieve nearly the best possible scalability and efficiency for a given set of client characteristics and desirable/feasible media quality. Furthermore, the results show that the new reliable protocols that transmit to each client at only twice the media play rate have similar performance to previous protocols that require clients to receive at many times the play rate.

This paper develops and validates an efficient analytical model for evaluating the performance of shared memory architectures with ILP processors. First, we instrument the SimOS simulator to measure the parameters for such a model and we find a surprisingly high degree of processor memory request heterogeneity in the workloads. Examining the model parameters provides insight into application behaviors and how they interact with the system. Second, we create a model that captures such heterogeneous processor behavior, which is important for analyzing memory system design tradeoffs. Highly bursty memory request traffic and lock contention are also modeled in a significantly more robust way than in previous work. With these features, the model is applicable to a wide range of architectures and applications. Although the features increase the model complexity, it is a useful design tool because the size of the model input parameter set remains manageable, and the model is still several orders of magnitude quicker to solve than detailed simulation. Validation results show that the model is highly accurate, producing heterogeneous per processor throughputs that are generally within 5 percent and, for the workloads validated, always within 13 percent of the values measured by detailed simulation with SimOS. Several examples illustrate applications of the model to studying architectural design issues and the interactions between the architecture and the application workloads.

This paper develops simple cost models for provisioning content distribution networks that use the simple and highly scalable bandwidth skimming protocol for streaming. New insight is obtained into: (1) how cost-effective proxy servers are in multicast streaming systems; (2) the most effective streaming protocol; and (3) the optimal proxy content, as a function of the system configuration and workload. A key result is that proxy servers are only cost effective if: (a) the origin server does not have a multicast capability; or (b) the file request rate is low, and thus multicast is not highly effective; or (c) the cost of a proxy server stream is a very small fraction (i.e., approximately 1/P) of the cost of an origin server stream, where P is the number of proxy servers and the cost of either type of stream includes both the server and network resource costs. For cases where proxy servers are cost effective, results in the paper provide the optimal proxy content and the most effective streaming protocol, as a function of a wide range of system configuration and workload parameters. In contrast to previous work, full file caching outperforms prefix caching over a significant region of this system design space, due to more efficient multicast streaming protocols as well as a more complete exploration of the practical system configuration space.

Recently proposed streaming protocols are able to deliver multimedia files on-demand with the required server bandwidth growing only logarithmically with the file request rate. The same efficiencies are achieved for network bandwidth if delivery is over a true broadcast channel. This paper considers the required network bandwidth for on-demand streaming over multicast delivery trees. We consider both simple canonical delivery trees, and more complex cases in which delivery trees are constructed using both existing and new algorithms for various randomly generated network topologies and client site locations. Our results quantify the potential savings from the use of multicast trees that are configured to minimize network bandwidth rather than the latency to the content server. Further, we show that it is possible to achieve reasonably close to the minimum possible bandwidth usage for both network and server simultaneously, with a practical on-demand streaming protocol.

Recently proposed scalable on-demand streaming protocols have previously been evaluated using a system cost measure termed the "required server bandwidth". For the scalable protocols that provide immediate service to each client when the server is not overloaded, this paper develops simple analytic models to evaluate two client-oriented quality of service metrics, namely (1) the mean client waiting time in systems where clients are willing to wait if a (well-provisioned) server is temporarily overloaded, and (2) the fraction of clients who balk (i.e., leave without receiving their requested media content) in systems where the clients will tolerate no or only very low service delays during a temporary overload. The models include novel approximate MVA techniques that appear to extend the range of applicability of customized AMVA to include questions focussed on state probabilities rather than on mean values, and to systems in which the operating points of interest do not include substantial client queues. For example, the new AMVA models accurately estimate the server bandwidth needed to achieve a balking rate as low as one in ten thousand. The analytic models can easily be applied to determine the server bandwidth needed for a given number of media files, anticipated total client request rate and file access frequencies, and target balking rate or mean wait. Results show that (a) scalable media servers that are configured with the "required server bandwidth" defined in previous work have low mean wait but may have unacceptably high client balking rates (i.e., greater than one in twenty), (b) for high to moderate client load, only a 10 - 50% increase in the previously defined required server bandwidth is needed to achieve a very low balking rate (e.g., one in ten thousand), and (c) media server performance (either mean wait or balking rate) degrades rapidly if the actual client load is more than 10% greater than the anticipated load.

This paper presents an extensive analysis of the client workloads for educational media servers at two major U.S. universities. The goals of the analysis include providing data for generating synthetic workloads, gaining insight into the design of streaming content distribution networks, and quantifying how much server bandwidth can be saved in interactive educational environments by using recently developed multicast streaming methods for stored content.

Inspired by recent techniques for reliable bulk data distribution, this paper develops scalable protocols for reliable on-demand delivery of streaming media. Models are developed that quantify the best possible scalability for given client characteristics. The results of the models are used to guide the design and assess the performance of the proposed streaming techniques. The new protocols, RPB and RBS, are relatively simple to implement and achieve nearly the best possible scalability and efficiency for a given set of client characteristics and desirable/feasible media quality.

Two recent techniques for multicast or broadcast delivery of streaming media can provide immediate service to each client request, yet achieve considerable client stream sharing which leads to significant server and network bandwidth savings. The paper considers: 1) how well these recently proposed techniques perform relative to each other and 2) whether there are new practical delivery techniques that can achieve better bandwidth savings than the previous techniques over a wide range of client request rates. The principal results are as follows: First, the recent partitioned dynamic skyscraper technique is adapted to provide immediate service to each client request more simply and directly than the original dynamic skyscraper method. Second, at moderate to high client request rates, the dynamic skyscraper method has required server bandwidth that is significantly lower than the recent optimized stream tapping/patching/controlled multicast technique. Third, the minimum required server bandwidth for any delivery technique that provides immediate real-time delivery to clients increases logarithmically (with constant factor equal to one) as a function of the client request arrival rate. Furthermore, it is (theoretically) possible to achieve very close to the minimum required server bandwidth if client receive bandwidth is equal to two times the data streaming rate and client storage capacity is sufficient for buffering data from shared streams. Finally, we propose a new practical delivery technique, called hierarchical multicast stream merging (HMSM), which has a required server bandwidth that is lower than the partitioned dynamic skyscraper and is reasonably close to the minimum achievable required server bandwidth over a wide range of client request rates.

Motivated by experience gained during the validation of a recent Approximate Mean Value Analysis (AMVA) model of modern shared memory architectures, this paper re-examines the “standard” AMVA approximation for non-exponential FCFS queues. We find that this approximation is often inaccurate for FCFS queues with high service time variability. For such queues, we propose and evaluate: (1) AMVA estimates of the mean residual service time at an arrival instant that are much more accurate than the standard AMVA estimate, (2) a new AMVA technique that provides a much more accurate estimate of mean center residence time than the standard AMVA estimate, and (3) a new AMVA technique for computing the mean residence time at a “downstream” queue which has a more bursty arrival process than is assumed in the standard AMVA equations. Together, these new techniques increase the range of applications to which AMVA may be fruitfully applied, so that for example, the memory system architecture of shared memory systems with complex modern processors can be analyzed with these computationally efficient methods.

The simplest video-on-demand (VOD) delivery policy is to allocate a new media delivery stream to each client request when it arrives. This policy has the desirable properties of “immediate service ” (there is minimal latency between the client request and

Disk caches are employed in distributed file systems to avoid network accesses at clients and to compensate for the speed differential between main memory and disk at file servers. Because of concerns about volatility, however, write requests have typically not benefitted from the presence of caches. Instead, they have been processed with some sort of write-through or periodic write-back approach to ensure the integrity of the stored data. The introduction of reasonably priced non-volatile (NV) memories has prompted interest in the use of such memory for write caching, at the server and/or at the client. This paper describes an investigation through trace-driven simulation experiments of several approaches to write caching in distributed systems, with both volatile and non-volatile caches. The results support the findings of earlier work that suggests important differences between caching in the traditional single-level caching environment and caching in a two-level caching environment. While policies focusing on temporal locality perform well for a single-level caching system, or at the client of a two-level caching system, they may not be suitable for use at the server in a two-level caching system. This is because locality characteristics in the reference stream seen at the server in a two-level caching system may be destroyed by caching at the client with a NV write cache large enough to hold the client's working set of dirty blocks. Policies focusing on amortizing the cost of a disk seek operation over multiple write-back operations perform better at the server of a two-level caching system.

No abstract available.

{sundaram,eager} Ocs.usask,ca Bus-based multiprocessors currently provide a cost-effective approach to achieving modestly parallel execution of large applrcatlons.Key (high level) parameters of such systems include the number of processors, the processor speed, the bus bandwidth, and the size of the caches.It is well-known that bus-based systems are not scalable in terms of the number of processors, as that parameter is tightly constrained by limitations of the bus architecture.As technology advances, however, the other key parameters (processor speed, bus bandwidth, cache size) will evolve, quite possibly to differing extents depending on the characteristics of the underlying technologies.This paper addresses the issue of how quickly bus bandwidth and cache size must scale, as processor speed scales, if bus-based multiprocessors are to be useful platforms for parallel computing in the future.For this purpose, we study the execution characteristics of a number of data parallel numerical and scientific applications, under a time-constrained scaling model, as system parameters are scaled.Our results indicate three types of applications.For Type I applications, parallel processing on future bus-based machines can be worthwhile only if either bus bandwidth scales as fast as processor speed, or if caches scale fast enough to eventually contain entire application data sets and appropriate affinity scheduling policies are employed, or a physically distributed memory architecture is employed.For Type II applications, parallel processing will be worthwhile regardless of the evolution of the memory system, assuming moderate scalings of bus bandwidth.For Type III applications, bus-based machines will become useless as a platform for parallel computing, unless the bus bandwidth scales as fast as the processor speed.

Multiprocessor interconnection networks based on low dimensional mesh or torus topologies and employing wormhole switching have become increasingly popular. Two concepts that have been proposed to improve the performance of such networks are Virtual Channel Flow Control (VCFC) and adaptive routing. Some previous studies have, however, found the latter technique to yield disappointing performance under uniform traffic patterns, while in this work we show that the former technique may result in degraded performance under certain non-uniform traffic patterns. This paper studies the interaction between VCFC and adaptive routing. It is shown that each of these techniques may yield unsatisfactory performance only when used in isolation, and that through their appropriate combination uniformly improved performance can be achieved. The consistent performance improvement and the magnitude of the performance improvement in some cases offers compelling motivation for implementing a combination of VCFC and adaptive routing in future machines. 1

Scheduling in a shared memory multiprocessor is often complicated by the fact that a unit of work may be processed more efficiently on one processor than on any other, due to factors such as the presence of required data in a local cache. The unit of work is said to have an "affinity" for the given processor, in such a case. The scheduling issue that has to be considered is the tradeoff between the goals of respecting processor affinities (so as to obtain improved efficiencies in execution) and of dynamically assigning each unit of work to whichever processor happens to be, at the time, least loaded (so as to obtain better load balance and decreased processor idle times). A specific context in which the above scheduling issue arises is that of shared memory multiprocessors with large per-processor caches or cached main memories. The shared-memory programming paradigm of such machines permits the dynamic scheduling of work. The data required by a unit of work may, however, often reside mostly in the cache of one particular processor, to which that unit of work thus has affinity. In this paper, two new "affinity scheduling" algorithms are proposed for a context in which the units of work have widely varying execution times. The two proposed algorithms are: (1) dynamic partitioned affinity scheduling and (2) wrapped partitioned affinity scheduling. An experimental study of these algorithms finds them to perform well in this context.>

Trace driven simulations were used to study the performance of several disk cache replacement policies for network file servers. It is shown that locality based approaches, such as the common least recently used (LRU) policy, which are known to work well on stand-alone disked workstations and at client workstations in distributed systems, are inappropriate at a fileserver. Quite simple frequency based approaches do better. More sophisticated frequency based policies (eg., that take into account the file type) may offer additional improvements.>

Parallel computing is increasingly important in the solution of large-scale numerical problems. The difficulty of efficiently hand-coding parallelism, and the limitations of parallelizing compilers, have nonetheless restricted its use by scientific programmers. In this paper we propose a new paradigm,chores, for the run-time support of parallel computing on shared-memory multiprocessors. We consider specifically uniform memory access shared-memory environments, although the chore paradigm should also be appropriate for use within the clusters of a large-scale nonuniform memory access machine. We argue that chore systems attain both the high efficiency of compiler approaches for the common case of data parallelism, and the flexibility and performance of user-level thread approaches for functional parallelism. These benefits are achieved within a single, simple conceptual model that almost entirely relieves the programmer and compiler from concerns of granularity, scheduling, and enforcement of synchronization constraints. Measurements of a prototype implementation demonstrate that the chore model can be supported more efficiently than can traditional approaches to either data or functional parallelism alone.

Consideration is given to the impact of hot-spot contention in distributed-memory multicomputer systems. Since the binary hypercube network offers a reasonable compromise between network cost and performance, two hypercube-based static interconnection networks are considered. One is a standard binary hypercube and the other is a binary hypercube-based hierarchical interconnection network. The impact of hot-spot contention in both these networks is studied.>

Spinning, or busy waiting, is commonly employed in parallel processors when threads of execution must wait for some event, such as synchronization with another thread. Because spinning is purely overhead, it is detrimental to both user response time and system throughput. The effects of two environmental factors, multiprogramming and data-dependent execution times, on spinning are discussed, and it is shown how the choice of scheduling discipline can be used to reduce the amount of spinning in each case.>

The influence of client and server cache sizes and the number of clients on caching performance is studied through trace-driven simulation. The results indicate that the locality of reference in disk block reference patterns allows relatively small caches to reduce significantly the number of disk accesses required. File server cache performance is significantly different from client cache performance owing to the capture of disk block references by the client caches. The major factor influencing overall miss ratio statistics (actual disk reference frequencies) is found to be the maximum of the server cache size and the size of client caches.>

Multiprocessors require an interconnection network to connect processors with memory modules. The performance of the interconnection network can have a large effect upon overall system performance, and, therefore, methods are needed to model and compare alternative network architectures.

A performance analysis of a class of hierarchical interconnection networks is presented. The analysis includes both static analysis (i.e. queueing delays are neglected) and queueing analysis. In both cases, the hierarchical networks are shown to have better cost-benefit ratios. The queueing analysis is also validated by several simulation experiments. The impact of two performance enhancement schemes-replication of links and improved routing algorithms-on hierarchical interconnection network performance is also presented.>

The tradeoff between speedup and efficiency that is inherent to a software system is investigated. The extent to which this tradeoff is determined by the average parallelism of the software system, as contrasted with other, more detailed, characterizations, is shown. The extent to which both speedup and efficiency can simultaneously be poor is bound: it is shown that for any software system and any number of processors, the sum of the average processor utilization (i.e. efficiency) and the attained fraction of the maximum possible speedup must exceed one. Bounds are given on speedup and efficiency, and on the incremental benefit and cost of allocating additional processors. An explicit formulation, as well as bounds, are given for the location of the knee of the execution time-efficiency profile, where the benefit per unit cost is maximized.>

Load sharing in a distributed system is the process of transparently sharing workload among the nodes in the system to achieve improved performance. In non-migratory load sharing, jobs may not be transferred once they have commenced execution. In load sharing with migration, on the other hand, jobs in execution may be interrupted, moved to other nodes, and then resumed.

Processor scheduling on multiprocessor systems that simultaneously run concurrent applications is currently not well-understood. This paper reports a preliminary investigation of a number of fundamental issues which are important in the context of scheduling concurrent jobs on multiprogrammed parallel systems. The major motivation for this research is to gain insight into system behaviour and understand the basic principles underlying the performance of scheduling strategies in such parallel systems. Based on abstract models of systems and scheduling disciplines, several high level issues that are important in this context have been analysed.

An algorithm for computing bounds on the performance measures of multiple-class, product-form queuing networks is presented. The algorithm offers the user a hierarchy of bounds with differing accuracy levels and computational cost requirements. Unlike previously proposed bounding algorithms, the algorithm is applicable to all of the types of product-form queuing networks that are commonly used in computer system and computer-communication network applications.

Rather than proposing a specific load sharing policy for implementation, the authors address the more fundamental question of the appropriate level of complexity for load sharing policies. It is shown that extremely simple adaptive load sharing policies, which collect very small amounts of system state information and which use this information in very simple ways, yield dramatic performance improvements. These policies in fact yield performance close to that expected from more complex policies whose viability is questionable. It is concluded that simple policies offer the greatest promise in practice, because of their combination of nearly optimal performance and inherent stability.

One goal of locally distributed systems is to facilitate resource sharing. Most current locally distributed systems, however, share primarily data, data storage devices, and output devices; there is little sharing of computational resources. Load sharing is the process of sharing computational resources by transparently distributing the system workload. System performance can be improved by transferring work from nodes that are heavily loaded to nodes that are lightly loaded.

In applications of queueing network models to computer system performance prediction, the computational effort required to obtain an exact equilibrium solution of a model may not be justified by the accuracy actually required.In these cases, there is a need for approximation or bounding techniques that can provide the necessary information with less computational effort.This paper presents a new technique that yields performance bounds for single-class separable queueing networks consisting of fixed-rate and delay service centers.Unlike previous approximation or bounding techniques, there is a smooth trade-off between computational effort and accuracy.Any level of accuracy (including the exact solution) can be guaranteed by investing the necessary computational effort.Performance bounds that are sufficiently tight for most practical purposes may be obtained with a fraction of the effort required for the exact solution.Since bounds are produced, as opposed to approximations, guarantees about the accuracy of a model solution can be provided.

The problem of concurrency control in distributed database systems in which site and communication link failures may occur is considered. The possible range of failures is not restricted; in particular, failures may induce an arbitrary network partitioning. It is desirable to attain a high “level of robustness” in such a system; that is, these failures should have only a small impact on system operation. A level of robustness termed maximal partial operability is identified. Under our models of concurrency control and robustness, this robustness level is the highest level attainable without significantly degrading performance. A basis for the implementation of maximal partial operability is presented. To illustrate its use, it is applied to a distributed locking concurrency control method and to a method that utilizes timestamps. When no failures are present, the robustness modifications for these methods induce no significant additional overhead.

Applications of queueing network models to computer system performance prediction typically involve the computation of their equilibrium solution. When numerous alternative systems are to be examined and the numbers of devices and customers are large, however, the expense of computing the exact solutions may not be warranted by the accuracy required. In such situations, it is desirable to be able to obtain bounds on the system solution with very little computation. Asymptotic bound analysis (ABA) is one technique for obtaining such bounds. In this paper, we introduce another bounding technique, called balanced job bounds (BJB), which is based on the analysis of systems in which all devices are equally utilized. These bounds are tighter than ABA bounds in many cases, but they are based on more restrictive assumptions (namely, those that lead to separable queueing network models).