Runtime Network-level Monitoring Framework in the Adaptation of Distributed Time-critical Cloud Applications

Many distributed time-critical applications have emerged on the Internet in recent decades, involving for example sensor-based early warning systems, online gaming and instant messaging. Such applications can be virtualised and distributed in a federated Cloud environment. Ensuring that these types of application are able to offer favourable service quality has been a challenging issue due to runtime variations in network conditions intrinsic to connections between individual application components replicated and distributed across different Cloud infrastructures. In this paper, we propose a lightweight method for performing network-level monitoring that can be used to guide the autonomous selection of optimal connections between running components, so improving distributed application performance at runtime. This solution contributes towards realising self-adaptation capabilities for time-critical applications by implementing a non-intrusive monitoring technique for key network-level parameters including round-trip time (RTT), packet loss, throughput and/or jitter. The experimental results show that the proposed framework has a low communication overhead and requires little processing power and memory capacity.