Michael J. Donahoo

Multiple-channel multicast scheduling for scalable bulk-data transport

A key technique for allowing servers to handle a large volume of requests for file transfers is to multicast the data to the set of requesting clients. Typically the paths from the server to the clients will be heterogeneous in bandwidth availability. Multiple-channel multicast (MCM) is an approach that can be used to handle this heterogeneity. In this approach the data is multicast over multiple channels, each addressed as a separate multicast group. Each receiver subscribes to a set of channels (i.e., joins the corresponding multicast groups) commensurate with its own rate capabilities. Of particular interest in the design of MCM schemes is the scheduling of data transmission across the multiple channels to accommodate asynchronous requests from clients. In this paper we present and analyze a new multiple-channel multicast approach called partition organization scheduling. The scheme is designed to result in good reception efficiency when compared to existing proposals while improving on their performance when other measures of interest are considered.

Scalable multicast representative member selection

In multicast applications requiring receiver feedback, the primary impediment to receiver-set scalability stems from the feedback implosion problem. Some applications only need feedback from a single, representative receiver. For example, an adaptive video server may only need to know the video quality for the worst receiver. Soliciting a response from a receiver with the specific metric value is difficult because a receiver does not know the metric value at other receivers and, therefore, cannot assess if it should report its metric value. Previous work in feedback aggregation solicits a response from any single receiver, not a receiver with some particular metric characteristic. In addition, existing solutions exhibit high application specificity, allowing very little protocol control. We propose adaptation of two existing approaches to representative selection: backoff timers and probabilistic polls. Both adapted approaches are application independent protocols and allow control over the definition of implosion, probability of implosion, and target metric values. We explicitly consider the inherent tradeoff between implosion and delay.

Aspect-driven, data-reflective and context-aware user interfaces design

The increasing use of Web-based applications continues to broaden the user groups of enterprise applications at large. Since ordinary users often equate the quality of user interface (UI) with the quality of the entire application, the importance of providing easy-to-use UIs has been significantly increasing. Unfortunately, designing a single UI satisfying all end users remains challenging. To address this issue, researchers and developers are looking to Context-aware/Adaptive UIs (CUIs) that aim to provide end users with more personalized user interaction experiences. Although multiple proposals have been made, very few production systems provide such malleable interfaces due to the excessive cost of development and maintenance. In this paper, we propose a technique that aims to reduce development and maintenance efforts of CUI to a level comparable with a single UI. Unlike most of the existing CUI approaches, our technique does not involve an external UI model. Instead, it aims to reflect runtime-information and structures already captured in the application, while extending them to provide an appropriate CUI. With this technique, developers do not design forms or tables directly for each page or panel. Instead they design generic and reusable transformation rules capable of presenting application data instances in the UI while considering the runtime context. To demonstrate our technique and its impact on CUI development and maintenance, we provide a case study. Moreover, we present our experience from its application to an existing production-level enterprise application, with high demands on performance.

Towards intelligent integration of heterogeneous information sources

Current methodologies for information integration are inadequate for solving the problem of integration of large scale, distributed information sources (e.g. databases, free-form text, simulation etc.). The existing approaches are either too restrictive and complicated as in the “federated” (global model) approach or do not provide the necessary functionality as in the “multidatabase” approach. We propose a hybrid approach combining the advantages of both the federated and multidatabase techniques which we believe provide the most feasible avenue for large scale integration. Under our architecture, the individual data site administrators provide anaugmented export schemaspecifying knowledge about the sources of data (where data exists), their structure (underlying data model or file structure), their content (what data exists), and their relationships (how the data relates to other information in its domain). The augmented export schema from each information source provides an intelligent agent, called the “mediator”, knowledge which can be used to infer information on some of the existing inter-system relationships. This knowledge can then be used to generate a partially integrated, global view of the data.

Grouping techniques for update propagation in intermittently connected databases

We consider an environment where one or more servers carry databases that are of interest to a community of clients. The clients are only intermittently connected to the server for brief periods of time. Clients carry a part of the database for their own processing and accumulate local updates while disconnected. We call this the Intermittently Connected Database (ICDB) environment. ICDBs have a wide variety of applications including sales force automation, insurance claim processing, and mobile workforces. Our focus is on the problem of update propagation at the server in ICDBs and the associated processing at the clients. The typical client-centric approach involves the communication and processing of updates and transactions on a per-client basis, ignoring the overlap of data between clients. The complexity of this approach is in the order of the number of connecting clients, thereby limiting the scalability of the server. We propose a data-centric approach which clusters data into groups and assigns to each client one or more of these groups. The proposed scheme results in server processing complexity on the order of the number of groups, which we control. We propose various techniques for grouping and discuss the processing required at the clients to enable the grouping approach. While the client-centric approach is expected to significantly degrade with the increasing number of clients, we expect that a properly designed grouping scheme will sustain a number of clients that is significantly larger. A prototype has been developed and performance studies are in progress.

On separation of platform-independent particles in user interfaces

The complexity of user interface (UI) design grows quickly with the number of application concerns. Such complexity compounds with additional requirement of contextual-awareness (i.e., adapt to user location, skill level, etc.) and support of heterogeneous devices and platforms (e.g., web, mobile app). Implementation support of such a wide-range of orthogonal concerns often results in restatement of a significant portion of the UI description using platform-specific components. Replication requires repeated implementation decision, greatly increasing development costs since each version/context variant may need separate development. Naturally, such replication also produces error prone maintenance because code updates must correlate among all replicas. Using separation of concerns, the application can be decomposed into fine-grain fragments, which we call particles, some of which are platform independent and others are not. Using this decomposition, this paper addresses the above inefficiency by dynamically composing particles at runtime that match user demands, context, and target platform.

Scaling replica maintenance in intermittently synchronized mobile databases

To avoid the high cost of continuous connectivity, a class of mobile applications employs replicas of shared data that are periodically updated. Updates to these replicas are typically performed on a client-by-client basis--that is, the server individually computes and transmits updates to each client--limiting scalability. By basing updates on replica groups (instead of clients), however, update generation complexity is no longer bound by client population size. Clients then download updates of pertinent groups. Proper group design reduces redundancies in server processing, disk usage and bandwidth usage, and dimininishes the tie between the complexity of updating replicas and the size of the client population. In this paper, we expand on previous work done on group design, include a detailed I/O cost model for update generation, and propose a heuristic-based greedy algorithm for group computation. Experimental results with an adapted commercial replication system demonstrate a significant increase in overall scalability over the client-centric approach.

Automated Business Rules Transformation into a Persistence Layer

Abstract Enterprise Information Systems maintain data with respect to various business processes. These processes consist of business operations restricted by business rules expressed as preconditions and post-conditions. Each rule must be considered and enforced throughout the system, from user interface to persistence storage. Such rule evaluation in multiple contexts results in both significant rule restatement and high maintenance complexity, as there is no single focal point for capturing and reusing these rules. In this paper, we apply the Aspect-Oriented Design Approach to the persistence layer to simplify business rules management, enforce business rules throughout the system and consequently decrease development and maintenance efforts. Our preliminary results show that it is possible to define business rules in a single place and then apply them automatically in a persistence layer. We retrieve data sets restricted by given operation post-conditions with respect to current execution context without any manual rule restatement. This paper provides a small case study emphasizing the benefits and future challenges.

Towards effective adaptive user interfaces design

The increasing use of Web-based applications continues to broaden the user groups of enterprise applications at large. The importance of providing easy-to-use user interfaces (UIs) that conform to each users specific preferences, such as different skill levels, capabilities and physical locations has, therefore, been significantly increasing. Unfortunately, designing a single UI satisfying all end users remains challenging. To address this issue, researchers and developer are looking to Adaptive User Interfaces (AUIs) that aim to provide end users with more personalized user interaction experiences. However, very few production system provide such malleable interfaces due to the excessive cost for the development and maintenance. In this paper, we propose a technique that provides AUIs for production enterprise systems while reducing development and maintenance efforts to a level comparable with a single UI development, called Rich Entity Aspect/Audit Design (READ). READ complies with application development standards used in industry to support an easy transition from design to production systems. We conclude by evaluating our approach along with a case study that demonstrates reduction in development and maintenance efforts while preserving performance.

Center selection and migration for wide-area multicast routing

Interdomain multicast routing is a challenge, requiring a scalable solution that is suitable for diverse application characteristics. Center-based routing trees overcome some of the limitations associated with traditional shortest-path tree routing, however they present a new set of routing problems. Speciically, the choice of the center router can be critical in determining the multicast routing performance. We present and assess algorithms to solve two problems: selection of an initial center router, and migration of the center router in response to dynamic changes in the group membership during the application lifetime. These two problems, and our proposed solutions, may have more general applicability in wide-area networking, e.g., in placing shared resources in the network or estimating bandwidth utilization for multicast distribution.