CLAM: cross-layer adaptation manager for service-based applications

Abstract

The heterogeneous and dynamic execution context of service-based applications (SBA) makes the problem of adaptation critical. However in most cases the adaptation is not trivial due to the following facts. First, SBA has a complex layered system where the application is implemented through a composition of services, which in turn are provided by platforms and run on top of infrastructures. Second, as a result of this multi-level application system there exist several adaptation approaches isolated from each other, which focus on a specific concern of one level ignoring the overall impact of the adaptation on the whole service-based system. To tackle this problem we propose a cross-layer adaptation manager (CLAM) whose contribution is two-fold: (i) It provides a platform that integrates and coordinates existing analysis and adaptation tools, which target specific system concerns, to assess the impact of an adaptation at the different levels. (ii) Covering the whole system for the SBA, it provides an analysis algorithm that incrementally constructs consistent adaptation strategies starting from an initial adaptation trigger originated at any level. The paper introduces the proposed approach and presents its first implementation with concrete analysis and adaptation tools.