William C. Cheng

A coordinated data collection approach: design, evaluation, and comparison

We consider the problem of collecting a large amount of data from several different hosts to a single destination in a wide-area network. This problem is important since improvements in data collection times in many applications such as wide-area upload applications, high-performance computing applications, and data mining applications are crucial to performance of those applications. Often, due to congestion conditions, the paths chosen by the network may have poor throughput. By choosing an alternate route at the application level, we may be able to obtain substantially faster completion time. This data collection problem is a nontrivial one because the issue is not only to avoid congested link(s), but to devise a coordinated transfer schedule which would afford maximum possible utilization of available network resources. Our approach for computing coordinated data collection schedules makes no assumptions about knowledge of the topology of the network or the capacity available on individual links of the network. This approach provides significant performance improvements under various degrees and types of network congestions. To show this, we give a comprehensive comparison study of the various approaches to the data collection problem which considers performance, robustness, and adaptation characteristics of the different data collection methods. The adaptation to network conditions characteristics are important as the above applications are long lasting, i.e., it is likely changes in network conditions will occur during the data transfer process. In general, our approach can be used for solving arbitrary data movement problems over the Internet. We use the Bistro platform to illustrate one application of our techniques.

Large-scale data collection: a coordinated approach

In this paper we consider the problem of collecting a large amount of data from several different hosts to a single destination in a wide-area network. Often, due to congestion conditions, the paths chosen by the network may have poor throughput. By choosing an alternate route at the application level, we may be able to obtain substantially faster completion time. This data collection problem is a nontrivial one because the issue is not only to avoid congested link(s), but to devise a coordinated transfer schedule which would afford maximum possible utilization of available network resources. In this paper we present an approach for computing coordinated data collection schedules, which can result in significant performance improvements. We make no assumptions about knowledge of the topology of the network or the capacity available on individual links of the network, i.e., we only use end-to-end information. Finally, we also study the shortcomings of this approach in terms of the gap between the theoretical formulation and the resulting data transfers in wide-area networks. In general, our approach can be used for solving arbitrary data movement problems over the Internet. We use the Bistro platform to illustrate one application of our techniques.

Bistro: a framework for building scalable wide-area Upload applications

Hot spots are a major obstacle to achieving scalability in the Internet. At the application layer, hot spots are usually caused by either (a) high demand for some data or (b) high demand for a certain service. This high demand for data or services, is typically the result of a real-life event involving availability of new data or approaching deadlines; therefore, relief of these hot spots may improve quality of life. At the application layer, hot spot problems have traditionally been dealt with using some combination of (1) increasing capacity; (2) spreading the load over time, space, or both; and (3) changing the workload.We note that the classes of solutions stated above have been studied mostly in the context of applications using the following types of communication (a) one-to-many, (b) many-to-many, and (c) one-to-one. However, to the best of our knowledge there is no existing work on making applications using many-to-one communication scalable and efficient (existing solutions, such as web based submissions, simply use many independent one-to-one transfers). This corresponds to an important class of applications, whose examples include the various upload applications such as submission of income tax forms, conference paper submission, proposal submission through the NSF FastLane system, homework and project submissions in distance education, voting in digital democracy applications, voting in interactive television, and many more. Consequently, the main focus of this paper is scalable infrastructure design for relief of hot spots in wide-area upload applications.The main contributions of this paper are as follows. We state (a) a new problem, specifically, the many-to-one communication, or upload, problem as well as (b) the (currently) fundamental obstacles to building scalable wide-area upload applications. We also propose a general framework, which we term the Bistro system, for a class of solutions to the upload problem. In addition, we suggest a number of open research problems, within this framework, throughout the paper.

Optimal routing for closed queueing networks

Abstract Optimal routing is an important subclass of resource allocation and load balancing problems that has applications in file allocation, distributed database systems, local area networks, etc. In this article, we present a generalization of the vertex allocation theorem of Tripathi and Woodside. The vertex allocation theorem applies to routing of single customer chains in closed product form networks. Basically, the theorem states that to maximize the sum of throughputs, each customer should consistently use the same server for each request type rather than probabilistically choose among alternatives for each request. We show that vertex allocation is valid for any network of quasi-reversible queues and that the objective function can be any Markov reward function on the state space of the queueing network model. In many optimization problems the customers are not all known a priori but rather enter the system dynamically. We address the problem of optimally choosing the routing for a new customer being added to an operational system. We model the system as a closed multichain queueing network and show that properties of networks of quasi-reversible queues can be used to obtain efficient optimal routing algorithms for this case.

Multipath streaming: optimization and evaluation

Quality of service (QoS) in delivery of continuous media (CM) over the Internet is still relatively poor, which is largely a result of packet losses often due to congestion. Previous work has shown that the use of multiple paths, existing in the network between a set of senders and a receiver, to deliver CM can be beneficial and should lead to improved QoS. In this paper, we study the problem of load distribution, i.e., properly distributing a CM stream among the multiple paths. We first focus on determining an appropriate optimization objective for computing the load distribution. We then conduct a performance study to understand the goodness of our optimization objectives and the resulting benefits of using multiple paths to deliver CM data.

Performance of batch-based digital signatures

A digital signature is an important type of authentication in a public-key (or asymmetric) cryptographic system, and it is in wide use. The performance of an Internet server computing digital signatures online is limited by the high cost of modular arithmetic. One simple way to improve the performance of the server is to reduce the number of computed digital signatures by combining a set of documents into a batch in a smart way and signing each batch only once. This reduces the demand on the CPU but requires extra information to be sent to clients. We investigate the performance characteristics of online digital signature batching schemes. We give a semi-Markov model of a gated batch-based digital signature server and its approximate solution. We validate the solutions of the analytical model through both emulation and simulation. Our study shows that significant computational benefits can be obtained from batching without significant increases in the amount of additional information that needs to be sent to the clients.

Bounding errors introduced by clustering of customers in closed product-form queuing networks

Product-form queuing network models have been widely used to model systems with shared resources such as computer systems (both centralized and distributed), communication networks, and flexible manufacturing systems. Closed multichain product-form networks are inherently more difficult to analyze than open networks, due to the effect of normalization. Results in workload characterization for closed networks in the literature are often for networks having special structures and only specific performance measures have been considered. In this article, we drive certain properties (insensitivity of conditional state probability distributions and fractional-linearity of Markov reward functions) for a broad class of closed multichain product-form networks. These properties are derived using the most basic flow balance conditions of product-form networks. Then we show how these basic properties can be applied in obtaining error bounds when similar customers are clustered together to speed up computation.

Performance study of online batch-based digital signature schemes

A digital signature is an important type of authentication in a public-key (or asymmetric) cryptographic system, and it is widely used in many digital government applications. We, however, note that the performance of an Internet server computing digital signatures online is limited by the high cost of modular arithmetic. One simple way to improve the performance of the server is to reduce the number of computed digital signatures by combining a set of documents into a batch in a smart way and signing each batch only once. This approach could reduce the demand on the CPU but require more network bandwidth of sending extra information to clients. In this paper, we investigate performance of different online digital signature batching schemes. That is, we provide a framework for studying as well as analyzing performance of a variety of such schemes. The results show that substantial computational benefits can be obtained from batching without significant increases in the amount of additional information that needs to be sent to the clients. Furthermore, we explore the potential benefits of considering more sophisticated batching schemes. The proposed analytical framework uses a semi-Markov model of a batch-based digital signature server. Through the emulation and the simulation, the results show the accuracy and effectiveness of our proposed analytic framework.

Towards global collaboration tools

The introduction of collaboration applications on the Internet has enriched communication between diverse users via chatting, and audio and video conferencing. It has also facilitated numerous real life tasks such as conducting of business meetings on the web, distance learning, and distributed authoring. Unfortunately, there are several difficulties that prevent collaboration applications from being ubiquitous on the Internet. The major difficulty is the lack of standards for communication between different collaboration applications which restricts collaboration to only instances of the same implementation, and in turn, limits the usability of such applications. In this paper, we propose a novel collaboration protocol for the exchange of information between users in collaborative environments. The proposed protocol is not tied to a specific implementation of collaboration tools which would enable several tools of the same type to communicate with each other independently of their implementation details. We also discuss a framework architecture for integration of several collaboration tools into one application. The proposed framework utilizes our proposed protocol, and adopts an HTTP-based communication model in order to overcome another difficulty which is possible firewalls constraints. In particular, we extend the standard HTTP streaming technology to accommodate several forms of collaboration.

A performance study of Bistro, a scalable upload architecture

Hot spots are a major obstacle to achieving scalability in the Internet. We have observed that the existence of hot spots in upload applications (whose examples include submission of income tax forms and conference paper submission) is largely due to approaching deadlines. The hot spot is exacerbated by the long transfer times. To address this problem, we proposed Bistro, a framework for building scalable wide-area upload applications, where we employ intermediaries, termed bistros, for improving the efficiency and scalability of uploads. Consequently, appropriate assignment of clients to bistros has a significant effect on the performance of upload applications and thus constitutes an important research problem. Therefore, in this paper we focus on the assignment of clients to bistros problem and present a performance study which demonstrates the potential performance gains of the Bistro framework.