Biblio

Moving target defense is an area of network security research in which machines are moved logically around a network in order to avoid detection. This is done by leveraging the immense size of the IPv6 address space and the statistical improbability of two machines selecting the same IPv6 address. This defensive technique forces a malicious actor to focus on the reconnaissance phase of their attack rather than focusing only on finding holes in a machine's static defenses. We have a current implementation of an IPv6 moving target defense entitled MT6D, which works well although is limited to functioning in a peer to peer scenario. As we push our research forward into client server networks, we must discover what the limits are in reference to the client server ratio. In our current implementation of a simple UDP echo server that binds large numbers of IPv6 addresses to the ethernet interface, we discover limits in both the number of addresses that we can successfully bind to an interface and the speed at which UDP requests can be successfully handled across a large number of bound interfaces.
 

Automated server parameter tuning is crucial to performance and availability of Internet applications hosted in cloud environments. It is challenging due to high dynamics and burstiness of workloads, multi-tier service architecture, and virtualized server infrastructure. In this paper, we investigate automated and agile server parameter tuning for maximizing effective throughput of multi-tier Internet applications. A recent study proposed a reinforcement learning based server parameter tuning approach for minimizing average response time of multi-tier applications. Reinforcement learning is a decision making process determining the parameter tuning direction based on trial-and-error, instead of quantitative values for agile parameter tuning. It relies on a predefined adjustment value for each tuning action. However it is nontrivial or even infeasible to find an optimal value under highly dynamic and bursty workloads. We design a neural fuzzy control based approach that combines the strengths of fast online learning and self-adaptiveness of neural networks and fuzzy control. Due to the model independence, it is robust to highly dynamic and bursty workloads. It is agile in server parameter tuning due to its quantitative control outputs. We implemented the new approach on a testbed of virtualized data center hosting RUBiS and WikiBench benchmark applications. Experimental results demonstrate that the new approach significantly outperforms the reinforcement learning based approach for both improving effective system throughput and minimizing average response time.
 

Monitoring is an important issue in cloud environments because it assures that acquired cloud slices attend the user's expectations. However, these environments are multitenant and dynamic, requiring automation techniques to offload cloud administrators. In a previous work, we proposed FlexACMS: a framework to automate monitoring configuration related to cloud slices using multiple monitoring solutions. In this work, we enhanced FlexACMS to allow dynamic and automatic attribution of monitoring configuration tasks to servers without administrator intervention, which was not available in previous version. FlexACMS also considers the monitoring server load when attributing configuration tasks, which allows load balancing between monitoring servers. The evaluation showed that enhancements reduced FlexACMS response time up to 60% in comparison to previous version. The scalability evaluation of enhanced version demonstrated the feasibility of our approach in large scale cloud environments.