Towards Hierarchical Management of Autonomic Components: A Case Study

Abstract

We address the issue of autonomic management in hierarchical component-based distributed systems. The long term aim is to provide a modelling framework for autonomic management in which QoS goals can be defined, plans for system adaptation described and proofs of achievement of goals by (sequences of) adaptations furnished. Here we present an early step on this path. We restrict our focus to skeleton-based systems in order to exploit their well-defined structure. The autonomic cycle is described using the Orc system orchestration language while the plans are presented as structural modifications together with associated costs and benefits. A case study is presented to illustrate the interaction of managers to maintain QoS goals for throughput under varying conditions of resource availability.

Integrating Autonomic Grid Components and Process-Driven Business Applications

“…Such more aggressive policies at the moment consist in varying the constant ∆ w that defines the number of workers to be added/removed when reconfiguring the parallelism degree of a BS. As shown in [17], the evaluation of such ∆ w can also take in account the overhead and the delay of the reconfiguration itself by using historical data. In the BS/GCM framework we are currently investigating the possibility to use a kind of exponential backoff increase/decrease protocol.…”

“…The corrective actions are usually taken from a library of known actions and the chosen action is determined by the result of the analysis phase. Finally, the actions planned are applied to the application during the execute phase [13,14,15,16,17,18].…”

Targeting Heterogeneous Architectures via Macro Data Flow

“…We are currently investigating the possibilities offered by several techniques, including the possibility to monitor execution of data parallel instructions on CPU cores/GPUs and to autonomically decide to schedule them on either GPU or CPU cores using Behavioural skeleton technology [23] and the possibility to use affinity scheduling and job stealing techniques to improve the memory usage in NUMA heterogeneous architectures.…”

Accessing Wireless Sensor Networks Via Dynamically Reconfigurable Interaction Models

“…The architecture of the Publish/Subscriber pattern, for instance, allows the reconfiguration of publishers, subscribers, and subscribed events. The Master/Slave pattern also allows the addition of new slaves to optimize task execution [19]; such is also incorporated in our solution within the context of a session. In spite of such reconfigurable system architecture definition, these works do not provide a session capturing an interaction's context, nor a pattern-based system evolution based on predefined rules conform to those pattern's semantics.…”