In this paper, we investigate the roles of replication vs. repair to achieve durability in large-scale distributed storage systems. Specifically, we address the fundamental questions: How does the lifetime of an object depend on the degree of replication and rate of repair, and how is lifetime maximized when there is a constraint on resources? In addition, in real systems, when a node becomes unavailable, there is uncertainty whether this is temporary or permanent; we analyze the use of timeouts as a mechanism to make this determination. Finally, we explore the importance of memory in repair mechanisms, and show that under certain cost conditions, memoryless systems, which are inherently less complex, perform just as well.

Pre-2018 CSE ID: CS2007-0900

We present a new Web middleware architecture that allows users to customize their view of the Web for optimal interaction and system operation when using non-traditional client machines such as wireless palmtops. Our Web Stream Customizers (WSC) are dynamically deployable and can be strategically located to achieve improvements in performance, reliability, or security. An important design feature is that Customizers provide two points of control in the communication path between client and server, supporting adaptive system-based and content-based customization. Our architecture exploits HTTP's proxy capabilities, allowing Customizers to be seamlessly integrated with the basic Web transaction model. We describe the WSC architecture and implementation, and illustrate its use with three non-trivial, adaptive Customizer applications that we have built. The overhead in our implementation which we present is small and tolerable, and we believe is outweighed by the benefits that Customizers provide.

Pre-2018 CSE ID: CS2001-0692

The concept of agents -- programs that are capable of transporting themselves across a [heterogeneous] network to execute and return results -- is a fascinating if troubled area of research. While theoretical advantages of agents have been well-established, few agent-based applications have been commercially successful. We argue that the lack of applications stems from a lack of understanding essential agent usage patterns. In this paper, we identify a set of fundamental patterns that support the design of agent-based applications that scale performance, reliability, and security. To evaluate their performance, some of these patterns were implemented in Java to demonstrate customizable and scalable performance.

Pre-2018 CSE ID: CS1999-0638

Virtual Network Computing, or VNC, is a popular thin client application used to access files and applications on remote computers. It is especially relevant as infrastructure to support ubiquitous computing applications, as it offers a way to run data-and-computation-intensive applications and allow users to access them through lightweight devices. However, VNC can suffer from significant losses in throughput when there is high latency between the client and server. In this work, we present a Message Accelerator proxy for VNC. This Message Accelerator mitigates high latency network effects while maintaining the advantages of a client-pull system. By operating near/on the server, it can send updates to the client at a rate corresponding to proxy-server interactions which are faster than client-server interactions. When testing using video, our Message Accelerator design results in frame rates an order of magnitude higher than plain VNC when running under high latency conditions.

Pre-2018 CSE ID: CS2009-0943

We present an architecture for Web Stream Customization (WSC) which allows users to customize their view of the Web for optimal interaction and system operation when using non-traditional client machines such as wireless palmtops. Our Web stream customizers are dynamically deployable and can be strategically located to achieve improvements in performance, reliability, or security. Customizers provide two points of control in the communication path between client and server, supporting adaptive system-based and content-based customization. Our architecture exploits HTTP's proxy capabilities, allowing Customizers to be seamlessly integrated with the basic Web transaction model. We describe the WSC architecture and its implementation, and illustrate its use with three non-trivial, adaptive Customizer applications that we have built. Our performance evaluation of the system shows that the overhead introduced is small and tolerable, and is outweighed by the benefits that Customizers provide.

Pre-2018 CSE ID: CS2001-0691

We introduce the notion of network flow buffering for direct HTTP streaming of web-based video clips, making it a more viable alternative to specialized streaming protocols, and in some cases making it a superior approach. A network flow buffer is simply a remote buffer that is dynamically deployed between a Web client and server, and that actively or passively regulates the flow of streaming video data. Network flow buffering is a simple approach to improving the performance of Web-based streaming video by buffering at a location where it can operate more effectively, providing benefits such as supplementing client buffering capacity, smoothing differences between WAN and LAN bandwidths, regulating data flow to a client, and supporting personalized caching to improve restart times. Network flow buffering is easily implemented as a Customizer in a middleware architecture we have developed (and reported on in WWW02) called the Web Stream Customizer Architecture (WSCA). The WSCA has a number of desirable features for supporting network flow buffering, including the dynamic deployment and relocation of software intermediaries between clients and servers that allow them to customize Web transactions, an extended usage model that supports customization of HTTP streaming and non-HTTP communication, and two points of control (one at the intermediary and one at or near the client) which can be used to supply feedback on conditions at the client to improve a network flow buffer▒s performance.

Pre-2018 CSE ID: CS2004-0776

Internet applications with wireless clients, which are rapidly increasing in popularity, face obstacles in terms of performance, reliability, and security. While much can be done to address these problems at the lower layers of the protocol stack, we focus on what can be done at the application level, which is complementary to the lower layer work but also has the advantage that it can be easily be deployed on top of existing or newer wireless Internet infrastructures. We propose that wireless Internet applications be enhanced through the use of a stylized form of mobile agent technology which we call behavior patterns. These are software templates that aid in the construction of these applications, and help in categorizing and constraining their behavior so that agent-hosts (the intermediary machines capable of executing mobile agents), are more willing to accept them for execution and provide them with the appropriate processing and communication resources. This paper provides an overview of the goals and design of this agent behavior pattern architecture, especially as it relates to supporting wireless Internet applications.

Pre-2018 CSE ID: CS2001-0693

Mobile Agents are a newly emerging network application technology in which programs can travel throughout a network to run on machines other than the originating host. Extensive research on Mobile Agents has been ongoing for many years and many Mobile Agent systems have been developed. However, there has not yet been any widespread acceptance of Mobile Agents as a viable technology by the Internet community. We examine the existing forms of mobile code technology and suggest using a limited form of Mobile Agent in conjunction with proxy servers for the purpose of client customization. We explore the issues of security, flexibility, deployability, and the utility of client customization using mobile agents and proxies, with the aim of developing a practical system that can be integrated into today's Internet. The benefits of using these agents for customization include performance improvements through the reduction of network latency, as well as improving the client's interface to distributed data sources through merging, filtering, and custom displays.

Pre-2018 CSE ID: CS2000-0641

Due to the lack of deployment of a network-layer multicast service, many overlay multicast protocols have been designed and deployed across the Internet to support content distribution. To our knowledge, however, none have provided a rigorous analysis of the problem or the effectiveness of their proposed solutions. Here, we set aside the engineering challenges of protocol design to focus on the fundamental graph problem. We begin by formulating the Overlay Network Content Distribution (OCD) problem and show that variants that attempt to optimize for either speed or bandwidth utilization are NP-complete. Using both a time-indexed Integer Program and a branch-and-bound search strategy, we calculate optimal solutions for small graphs. While solutions to OCD provide performance bounds, realistic deployment scenarios will not have global network information. Hence, we introduce an on-line variant, the Local-knowledge Overlay Content Distribution (LOCD) problem and show that no constant-competitive approximation exists. Instead, we present several heuristics that perform well in realistic topologies. We conclude with an evaluation of our global and local heuristics and enumerate a number of open problems.

Pre-2018 CSE ID: CS2005-0824