The design and performance of component middleware for QoS-enabled deployment and configuration of DRE systems

Abstract

Quality of service QoS-enabled component middleware can help reduce the complexity of deploying and con guring QoS aspects, such as priorities and rates of invocation. Few empirical studies have been conducted, however, to guide developers of distributed r e al-time and embedded DRE systems in choosing among alternative designs and performance optimizations. Moreover, few empirical studies have been conducted to examine the performance and exibility trade-o s between standardsbased and domain-speci c DRE middleware solutions. This paper makes three key contributions to research on QoS-enabled component middleware for DRE systems. First, it describes optimizations applied to an implementation of the OMG's Deployment and Con guration D&C of Components speci cation that enable performance trade-o s between QoS aspects of DRE systems. Second, it compares the performance of several dynamic and static con guration mechanisms to help guide the selection of suitable con guration mechanisms based on speci c DRE system requirements. Third, it compares the performance of our static standards-based approach to an avionics domain-speci c approach. Our results show that these optimizations 1 provide developers improved c ontrol over key trade-o s between exibility and performance at di erent stages of the DRE system lifecycle, 2 enhance trustworthiness of component-based DRE systems by supporting greater customization of how they are c on gured t o m e et speci c requirements of each application, and 3 o er greater exibility at a reasonable performance cost, compared to a domain-speci c approach.

Efficient and deterministic application deployment in component-based enterprise distributed real-time and embedded systems

“…The start launch phase of deployment produces the largest amount of jitter in the LE-DAnCE deployment process. Prior experiments [32] conducted on DAnCE showed this jitter stemmed from the dynamic loading of component implementations at runtime, which can be alleviated by directly compiling component implementations and plan metadata into the deployment infrastructure. While this approach reduces jitter and latency, it is also invasive to the D&C implementation, hard to maintain, and removes too much flexibility from the D&C toolchain.…”

Predictable deployment in component-based enterprise distributed real-time and embedded systems

“…The start launch phase of deployment produces the largest amount of jitter in the LE-DAnCE deployment process. Prior experiments [23] conducted on DAnCE showed this jitter stemmed from the dynamic loading of component implementations at runtime and can be alleviated by directly compiling component implementations and plan metadata into the deployment infrastructure [24]. While this approach reduces jitter and latency, it is also invasive to the D&C implementation, hard to maintain, and removes much of the flexibility from the D&C toolchain.…”

Minimum disruption service composition and recovery in mobile ad hoc networks

“…To characterize the structure of distributed applications that are expected to run in the mobile computing environments, we apply a component-based software model (2). All application components are constructed as autonomous services that perform independent operations (such as transformation and filtering) on the data stream passing through them.…”

“…These diverse application needs have fueled an increasing demand for new functionalities and services. To meet these demands, component-based software development (2) has been used to ensure the flexibility and maintainability of software systems. Service composition (3; 4; 5) is a promising technique for integrating loosely-coupled distributed service components into a composite service that provides end users with coordinated functionality, such as web services and multimedia applications.…”

“…For example, the reliability metric helps guide and evaluate the design of many ad hoc routing algorithms (10; 11) and component deployment mechanisms (12) using the path with maximum reliability for data/service delivery. There are two problems, however, with using reliability as a metric for service composition and recovery design: (1) it does not account for service repair and recovery and (2) reliability is a dynamic metric that is usually estimated based on the signal strength of a wireless link or the packet loss ratio along a path. Its constantly changing value may cause repeated service adjustments, especially if an application wants to use the path with maximum reliability.…”