Improving Availability in Large, Distributed Component-Based Systems Via Redeployment

Abstract

In distributed and mobile environments, the connections among the hosts on which a software system is running are often unstable. As a result of connectivity losses, the overall availability of the system decreases.

A Planning Method for Component Placement in Smart Item Environments Using Heuristic Search

“…Special focus is put on the evaluation of planning algorithms, e.g. an approximative algorithm based on ordered lists of hosts and components [8], a decentralised algorithm with an auctioning mechanism [9], greedy and clustering algorithms [11]. The input model allows specifying memory constraints, and evaluation of bandwidth constraints through frequency of message exchanges between components and the average message size.…”

“…Availability is defined as the degree to which a system or component is operational and accessible when it is required for use [6]. In a distributed component-based system, the availability depends on the placement of components [7,8,9] and hence availability is also a relevant evaluation criterion for deployment plans. Intuitively, the availability increases when the amount of data to be transferred over unreliable connections decreases.…”

Support for disconnected operation via architectural self-reconfiguration

“…Figure 2 shows a formal definition of the problem. An in-depth discussion of this problem, its generalization, and a detailed overview of possible approximation techniques is given in [11].…”

“…These algorithms require the data obtained during the monitoring stage. A detailed performance comparison of these algorithms is given in [11].…”

“…Similarly, selecting the best software component for assignment to a host is performed by choosing the component with the highest frequency of interaction with other components in the system, and the lowest required memory. Our algorithm automatically assigns different calibration factors to each one of these parameters, as detailed in [11]. The calibration factors denote the degree of "contribution" of each parameter to the selection of the best host and component.…”