A mobility-enabled architecture is usually required to facilitate the access of enterprise systems from mobile devices. Typically, mobility enabled architectures are separated in online architectures and offline architectures. The online architectures enable mobile clients to use the online model in which mobile clients access business logic and data located on backend systems using remote invocations. The offline architectures enable mobile clients to use the offline model in which mobile clients access business data locally on devices, and periodically synchronize data with backend systems. The selection of the architecture is based on the needs of the scenarios in accordance with the user requirements. For an enterprise system, there is usually more than one mobile scenario, and hence both online and offline architectures might be required. In this case, the discrepancies between the two architectures will increase the cost and decrease the interoperability between different solutions. In order to solve these problems, this paper proposes an innovative and lexible mobility enabled architecture that can be configured to be an online, an offline, or a mixed architecture in a unified framework. Using this architecture, different types (online, offline, or mixed) of mobile client applications could be built for different scenarios according to the user requirements. These applications can share the same mid-tier framework and access the same backend systems. Moreover, in this architecture, a unified access model to business logic is proposed to make it possible to adapt in time a mobile application from the online model to the offline model without architectural changes and vice versa. This paper introduces the basic requirements to build such an architecture, defines basic modules of the architecture and shows how to adapt this architecture to different (online, offline, or mixed) architectures using different configurations.

Reasoning about the correctness of a replication algorithm is a difficult endeavor. If correctness has to be shown for a component based architecture where a client request can lead to execution across different components or tiers, this is even more difficult. Existing formalisms are either restricted to systems with only one component, or make strong assumptions about the setup of the system. In this paper, we present a flexible framework to reason about exactly-once execution in a failure-prone replicated component based system. Our approach allows us to reason about the execution across the entire system, e.g., application server and database tier. If a given replication algorithm makes assumptions about some of the components, then those can be easily integrated into the reasoning process.

The increase in the number of mobile device users has led to an increased demand for mobile business applications. One of the main challenges in developing mobile business applications is to efficiently process business data in the resource constrained environment of mobile phones. Current approaches are seriously limited by the non-guaranteed quality of network connections or the restricted processing capabilities of mobile phones. We propose a lightweight asynchronous online framework (LAOF) that merges the benefits of both online and offline models. In our approach, business data is partially and lazily replicated to mobile phones. The mobile application executes read-only operations locally and tentatively simulates local responses to write operations before asynchronously executing on the remote server. The LAOF improves the traditional lazy replication approach by 1) efficiently compressing business data for mobile applications, 2) separating the consistency requirements for responses and business data, and 3) providing an approach that guarantees causal dependency for the asynchronous execution of write operations on the server side. In addition, we also demonstrate a sample application based on this approach and use it to evaluate the framework.

A growing number of mobile users prefer to access information and even make critical business decisions using their mobile phones. We present an integrated architecture developed at SAP Research as a platform for mobile business applications. The immediate goal is to allow mobile users to access information and services provided by the business back-end servers. The long-term goal is to develop techniques to overcome many problems mobile users face under such constraints as limited bandwidth, computational power and storage resources. The framework that we propose in this paper enables business applications connected to SAP through common mobile devices. SAP users can access business data and applications on the back- end server and execute tasks without a PC. We present components of the Lightweight Architecture (LWA) with specific components in the areas of data management, data compression, connectivity, and user interfaces.

This paper provides a brief description of a research project on using Ajax to enhance mobile Web applications for enterprise use. The project, known as MobileWeaver Ajax Framework, leverages enterprise SOA (Service Oriented Architecture) and the latest web technologies on mobile devices.

The goal of this poster is to describe our implementation of a newarchitecture enabling efficient integration between mobile phoneapplications and Web Services. Using this architecture, we haveimplemented a mobile shopping assistant described further. In orderto build this architecture, we designed an innovative XMLcompression mechanism to facilitate data exchange between mobilephones and Web Services. We also designed a smart connection managerto control asynchronous communication for all possible channels of amobile phone. In addition, we used diverse input modes in order toextend users' access to Web Services.

Replication is widely used in application server products to tolerate faults. An important challenge is to correctly coordinate replication and transaction execution for stateful application servers. Many current solutions assume that a single client request generates exactly one transaction at the server. However, it is quite common that several client requests are encapsulated within one server transaction or that a single client request can initiate several server transactions. In this paper, we propose a replication tool that is able to handle these variations in request/transaction association. We have integrated our approach into the J2EE application server JBoss. Our evaluation using the ECPerf benchmark shows a low overhead of the approach.