Performability of disk-array-based video servers

Video services are likely to dominate the traffic in future broadband networks. Most of these services will be provided by large- scale public-access video servers. Research to date has shown that disk arrays are a promising technology for providing the storage and throughput required to serve many independent video streams to a large customer population. Large disk arrays, however, are susceptible to disk failures which can greatly affect their reliability. In this paper, we discuss suitable redundancy mechanisms to increase the reliability of disk arrays and compare the performance of the RAID-3 and RAID-5 redundancy schemes. We use cost and performability analyses to rigorously compare the two schemes over a variety of conditions. Accurate cost models are developed and Markov reward models (with time-dependent reward structures) are developed and used to give insight into the tradeoffs between system cost and revenue earning potential. The paper concludes that for large-scale video servers, coarse-grained striping in a RAID-5 style of disk array is most cost effective.     

Some simple distributed algorithms for sparse networks

Summary. We give simple, deterministic, distributed algorithms for computing maximal matchings, maximal independent sets and colourings. We show that edge colourings with at most colours, and maximal matchings can be computed within deterministic rounds, where is the maximum degree of the network. We also show how to find maximal independent sets and -vertex colourings within deterministic rounds. All hidden constants are very small and the algorithms are very simple. Received: August 2000 / Accepted: November 2000     

Disk placement for arbitrary-rate playback in an interactive video server

Multimedia data, especially continuous media including video and audio objects, represent a rich and natural stimulus for humans, but require large amount of storage capacity and real-time processing. In this paper, we describe how to organize video data efficiently on multiple disks in order to support arbitrary-rate playback requested by different users independently. Our approach is to segment and decluster video objects and to place the segments in multiple disks using a restricted round-robin scheme, called prime round-robin (PRR). Its placement scheme provides uniform load balance of disks for arbitrary retrieval rate as well as normal playback, since it eliminates hot spots. Moreover, it does not require any additional disk bandwidth to support VCR-like operations such as fast-forward and rewind. We have studied the various effects of placement and retrieval schemes in a storage server by simulation. The results show that PRR offers even disk accesses, and the failure in reading segment by deadline occurs only at the beginning of new operations. In addition, the number of users admitted is not decreased, regardless of arbitrary-rate playback requests.     

An experimental analysis of digital video library servers

Much work on video servers has concentrated on movies on demand, in which a relatively small number of titles are viewed and users are given basic VCR-style controls. This paper concentrates on analyzing video server performance for non-linear access applications. In particular, we study two non-linear video applications: video libraries, in which users select from a large collection of videos and may be interested in viewing only a small part of the title; and video walk-throughs, in which users can move through an image-mapped representation of a space. We present a characterization of the workloads of these applications. Our simulation studies show that video server architectures developed for movies on demand can be adapted to video library usage, though caching is less effective and the server can support a smaller user population for non-linear video applications. We also show that video walk-throughs require extremely large amounts of RAM buffering to provide adequate performance for even a small number of users.     

Efficient support for interactive operations in multi-resolution video servers

In this paper, we present a placement algorithm that interleaves multi-resolution video streams on a disk array and enables a video server to efficiently support playback of these streams at different resolution levels. We then combine this placement algorithm with a scalable compression technique to efficiently support interactive scan operations (i.e., fast-forward and rewind). We present an analytical model for evaluating the impact of the scan operations on the performance of disk-arr ay-based servers. Our experiments demonstrate that: (1) employing our placement algorithm substantially reduces seek and rotational latency overhead during playback, and (2) exploiting the characteristics of video streams and human perceptual tolerances enables a server to support interactive scan operations without any additional overhead.     

Analysis,management and indexing of distributed multimedia performances based on audience feedback

Interactive performance systems allow multiple, networked users to take part in a performance either as players or audience members and influence the performance's development in real time. Among the most prominent research issues concerning the development of these systems are the provision of effective interaction capabilities and adequate means of expression to the audience. This article describes a set of methods for the detection, analysis, synchronization and rendering of audience behavior in real-time during such an event. Furthermore, it describes ways in which audience feedback is utilized to dynamically index the contents of a performance so as to embed this event in large-scale multimedia services. All these methods have been applied in the creation of MISSION, a multi-user game involving both players and audience. This article presents and analyzes the design of the system and the results of user trials with it.     

Traffic characterization algorithms for VBR video in multimedia networks

A network that offers deterministic, i.e., worst case, quality-of-service guarantees to variable-bit-rate (VBR) video must provide a resource reservation mechanism that allocates bandwidth, buffer space, and other resources for each video stream. Such a resource reservation scheme must be carefully designed, otherwise network resources are wasted. A key component for the design of a resource reservation scheme is the traffic characterization method that specifies the traffic arrivals on a video stream. The traffic characterization should accurately describe the actual arrivals, so that a large number of streams can be supported; but it must also map directly into efficient traffic-policing mechanisms that monitor arrivals on each stream. In this study, we present a fast and accurate traffic characterization method for stored VBR video in networks with a deterministic service. We use this approximation to obtain a traffic characterization that can be efficiently policed by a small number of leaky buckets. We present a case study where we apply our characterization method to networks that employ a dynamic resource reservation scheme with renegotiation. We use traces from a set of 25–30-min MPEG sequences to evaluate our method against other characterization schemes from the literature.     

Impact of QOS requirements on video coding for ATM networks

The broadband integrated services digital networks (B-ISDN) based on asynchronous transfer mode (ATM) technology can support a wide range of applications such as voice, video, still images, and data. Compression techniques increase the effective bandwidth utilization, but the bursty and asynchronous nature of the traffic can still lead to congestion in the network, and degradation of image quality and quality of service (QOS). Some of the features to provide better coding schemes for ATM networks are layered coding, resynchronization, buffering, interleaved schemes, constrained bit rate due to buffers, encapsulation with the RTP or AAL1 for clock recovery, lapped transforms, motion compensation, and optimal bit allocation for coders based on wavelet transforms. We review various techniques forimage and video coding such as transforms, motion compensation, vector quantization, and subband coding. We outline the impact of the cell loss ratio (CLR), delay and cell delay variation (CDV) on video coding: blocking effects, loss of frame synchronization, motion vectors, and vector quantization codewords. The open problems include tuning coding parameters to the available QOS provided by the network.     

Causality and atomicity in distributed computations

Summary. In a distributed system, high-level actions can be modeled by nonatomic events. This paper proposes causality relations between distributed nonatomic events and provides efficient testing conditions for the relations. The relations provide a fine-grained granularity to specify causality relations between distributed nonatomic events. The set of relations between nonatomic events is complete in first-order predicate logic, using only the causality relation between atomic events. For a pair of distributed nonatomic events X and Y, the evaluation of any of the causality relations requires integer comparisons, where and , respectively, are the number of nodes on which the two nonatomic events X and Y occur. In this paper, we show that this polynomial complexity of evaluation can by simplified to a linear complexity using properties of partial orders. Specifically, we show that most relations can be evaluated in integer comparisons, some in integer comparisons, and the others in integer comparisons. During the derivation of the efficient testing conditions, we also define special system execution prefixes associated with distributed nonatomic events and examine their knowledge-theoretic significance. Received: July 1997 / Accepted: May 1998     

Transformation and exchange of multimedia objects in distributed multimedia systems

One of the challenges in the design of a distributed multimedia system is devising suitable specification models for various schemas in different levels of the system. Another important research issue is the integration and synchronization of heterogeneous multimedia objects. In this paper, we present our models for multimedia schemas and transformation algorithms. They transform high-level multimedia objects into schemas that can be used to support the presentation and communication of the multimedia objects. A key module in the system is the Object Exchange Manager (OEM). In this paper, we present the design and implementation of the OEM module, and discuss in detail the interaction between the OEM and other modules in a distributed multimedia system.     

The BG distributed simulation algorithm

We present a shared memory algorithm that allows a set of f+1 processes to wait-free “simulate” a larger system of n processes, that may also exhibit up to f stopping failures. Applying this simulation algorithm to the k-set-agreement problem enables conversion of an arbitrary k-fault-tolerant{\it n}-process solution for the k-set-agreement problem into a wait-free k+1-process solution for the same problem. Since the k+1-processk-set-agreement problem has been shown to have no wait-free solution [5,18,26], this transformation implies that there is no k-fault-tolerant solution to the n-process k-set-agreement problem, for any n. More generally, the algorithm satisfies the requirements of a fault-tolerant distributed simulation.\/ The distributed simulation implements a notion of fault-tolerant reducibility\/ between decision problems. This paper defines these notions and gives examples of their application to fundamental distributed computing problems. The algorithm is presented and verified in terms of I/O automata. The presentation has a great deal of interesting modularity, expressed by I/O automaton composition and both forward and backward simulation relations. Composition is used to include a safe agreement\/ module as a subroutine. Forward and backward simulation relations are used to view the algorithm as implementing a multi-try snapshot\/ strategy. The main algorithm works in snapshot shared memory systems; a simple modification of the algorithm that works in read/write shared memory systems is also presented. Received: February 2001 / Accepted: February 2001     

Using rotational mirrored declustering for replica placement in a disk-array-based video server

In a video-on-demand (VOD) environment, disk arrays are often used to support the disk bandwidth requirement. This can pose serious problems on available disk bandwidth upon disk failure. In this paper, we explore the approach of replicating frequently accessed movies to provide high data bandwidth and fault tolerance required in a disk-array-based video server. An isochronous continuous video stream imposes different requirements from a random access pattern on databases or files. Explicitly, we propose a new replica placement method, called rotational mirrored declustering (RMD), to support high data availability for disk arrays in a VOD environment. In essence, RMD is similar to the conventional mirrored declustering in that replicas are stored in different disk arrays. However, it is different from the latter in that the replica placements in different disk arrays under RMD are properly rotated. Combining the merits of prior chained and mirrored declustering methods, RMD is particularly suitable for storing multiple movie copies to support VOD applications. To assess the performance of RMD, we conduct a series of experiments by emulating the storage and delivery of movies in a VOD system. Our results show that RMD consistently outperforms the conventional methods in terms of load-balancing and fault-tolerance capability after disk failure, and is deemed a viable approach to supporting replica placement in a disk-array-based video server.     

Deadlock detection in distributed database systems: a new algorithm and a comparative performance analysis

This paper attempts a comprehensive study of deadlock detection in distributed database systems. First, the two predominant deadlock models in these systems and the four different distributed deadlock detection approaches are discussed. Afterwards, a new deadlock detection algorithm is presented. The algorithm is based on dynamically creating deadlock detection agents (DDAs), each being responsible for detecting deadlocks in one connected component of the global wait-for-graph (WFG). The DDA scheme is a “self-tuning” system: after an initial warm-up phase, dedicated DDAs will be formed for “centers of locality”, i.e., parts of the system where many conflicts occur. A dynamic shift in locality of the distributed system will be responded to by automatically creating new DDAs while the obsolete ones terminate. In this paper, we also compare the most competitive representative of each class of algorithms suitable for distributed database systems based on a simulation model, and point out their relative strengths and weaknesses. The extensive experiments we carried out indicate that our newly proposed deadlock detection algorithm outperforms the other algorithms in the vast majority of configurations and workloads and, in contrast to all other algorithms, is very robust with respect to differing load and access profiles. Received December 4, 1997 / Accepted February 2, 1999     

Packet video transport in ATM networks with single-bit feedback

The promise of a broadband integrated service digital network has led to the design of mechanisms for efficient transport of real-time compressed video in packet switching networks. We examine feedback control for video transport in ATM networks where the available feedback is a single bit of information carried in the cell header. We investigate the performance of three single-bit schemes for source rate adaptation. Two were originally for congestion control of bursty data traffic and are modified to control video traffic. The third scheme conveys more information about the state of queue(s) at the bottleneck. The simulation results show that all three schemes for feedback control of VBR video streams work remarkably well. During severe network congestion, the signal quality degrades gracefully, but not uniformly across all connections. Based on insights from the initial simulations, we propose a scheme to improve the fairness of service and demonstrate its effectiveness.     

Vector time and causality among abstract events in distributed computations

An important problem in analyzing distributed computations is the amount of information. In event-based models, even for simple applications, the number of events is large and the causal structure is complex. Event abstraction can be used to reduce the apparent complexity of a distributed computation. This paper discusses one important aspect of event abstraction: causality among abstract events. Following Lamport [24], two causality relations are defined on abstract events, called weak and strong precedence. A general theoretical framework based on logical vector time is developed in which several meaningful timestamps for abstract events are derived. These timestamps can be used to efficiently determine causal relationships between arbitrary abstract events. The class of convex abstract events is identified as a subclass of abstract events that is general enough to be widely applicable and restricted enough to simplify timestamping schemes used for characterizing weak precedence. We explain why such a simplification seems not possible for strong precedence. Received: February 1994 / Accepted: July 1997     

MING-I: a distributed interactive multimedia document development mechanism

The next generation of interactive multimedia documents can contain both static media, e.g., text, graph, image, and continuous media, e.g., audio and video, and can provide user interactions in distributed environments. However, the temporal information of multimedia documents cannot be described using traditional document structures, e.g., Open Document Architecture (ODA) and Standard Generalized Mark-up Language (SGML); the continuous transmission of media units also raises some new synchronization problems, which have not been met before, for processing user interactions. Thus, developing a distributed interactive multimedia document system should resolve the issues of document model, presentation control architecture, and control scheme. In this paper, we (i) propose a new multimedia document model that contains the logical structure, the layout structure, and the temporal structure to formally describe multimedia documents, and (ii) point out main interaction-based synchronization problems, and propose a control architecture and a token-based control scheme to solve these interaction-based synchronization problems. Based on the proposed document model, control architecture, and control scheme, a distributed interactive multimedia document development mechanism, which is called MING-I, is developed on SUN workstations.     

Searching and editing MPEG-compressed video in a distributed online environment

WebClip (on-line demo at http://www.ctr.columbia.edu/webclip) is a compressed video searching and editing system operating over the World Wide Web. WebClip uses a distributed client-server model including a server engine for content analysis/editing, and clients for interactive controls of video browsing/editing. It specializes several unique features, including compressed-domain video feature extraction and manipulation, multi-resolution video access, content based video browsing/retrieval, and a distributed network architecture.     

Synchronization of distributed multimedia systems with user interactions

Without user interactions, multimedia presentations are just fancy slide shows with sound and video supports. User interactions by themselves do not change the temporal relationships among multimedia objects, such as texts, graphics, images, audio, and video, but affect the playback schedules. In this paper, we propose a synchronization mechanism to guarantee the quality of multimedia presentation with user interactions. In our protocol, each presentation site requests media transmission from the required media servers at certain time intervals prior to the playback deadlines, where these time intervals are the response times to cover possible experienced end-to-end delays and packet losses, and waits for an initial setup time to ensure intermedia synchronization before starting the presentation. Users may interact with the presentation. This synchronization mechanism solves the problems incurred by user interactions by determining the new presentation scenario produced by the interactive operation, calculating the corresponding setup time, and then rendering the new playback and retrieval schedules.     

A sound and complete algorithm for distributed commerce transactions

Summary. In a multi-party transaction (also called a distributed commerce transaction) agents face risks from dealing with untrusted agents. These risks are compounded in the face of deadlines, e.g., an agent may fail to deliver purchased goods by the time the goods are needed. We characterize the risks, and present a distributed algorithm that mitigates these risks, by using pairwise exchanges and trusted intermediaries. The algorithm generates a safe sequence of actions that completes a commerce transaction without risk, if such a sequence exists. We show that the algorithm is sound (produces only safe multi-agent action sequences) and complete (finds a safe sequence whenever one exists). The initial restriction of guaranteeing safety even when none of the principals trusts another can be relaxed in some cases, so we show how to handle principals that do trust each other and interact directly rather than through a trusted intermediary. Received: September 1997 / Accepted: December 1998     

Deterministic broadcasting in ad hoc radio networks

We consider the problem of distributed deterministic broadcasting in radio networks of unknown topology and size. The network is synchronous. If a node u can be reached from two nodes which send messages in the same round, none of the messages is received by u. Such messages block each other and node u either hears the noise of interference of messages, enabling it to detect a collision, or does not hear anything at all, depending on the model. We assume that nodes know neither the topology nor the size of the network, nor even their immediate neighborhood. The initial knowledge of every node is limited to its own label. Such networks are called ad hoc multi-hop networks. We study the time of deterministic broadcasting under this scenario. For the model without collision detection, we develop a linear-time broadcasting algorithm for symmetric graphs, which is optimal, and an algorithm for arbitrary n-node graphs, working in time . Next we show that broadcasting with acknowledgement is not possible in this model at all. For the model with collision detection, we develop efficient algorithms for broadcasting and for acknowledged broadcasting in strongly connected graphs. Received: January 2000 / Accepted: June 2001