BDMaaS+: Business-Driven and Simulation-Based Optimization of IT Services in the Hybrid Cloud

Abstract

The maturity of heterogeneous and hybrid public Cloud environments enables service providers to deploy there their complex IT services trusting these large and complex infrastructures. At the same time, evaluating the impact of changes at service configuration before and at the runtime is still a very challenging and difficult task. Moreover, a comprehensive performance evaluation of IT service configurations should not be limited just to costs for IT resource acquisition, but also include risk related elements such as Service Level Agreement (SLA) violation penalties and other intangibles. To support IT service providers in this difficult task, we developed Business-Driven Management as a Service Plus (BDMaaS+), a novel decision support tool that can evaluate IT service configuration through simulation with realistic service and network models. By allowing service providers to define expanded operational parameters, BDMaaS+ also enables what-if scenario analysis, thereby opening interesting possibilities at the planning level. Experimental results, collected from our thorough evaluations, demonstrate how a service provider can leverage BDMaaS+ to explore the potential of high-level business SLA changes and data center additions.

Reinforcement Learning vs. Computational Intelligence: Comparing Service Management Approaches for the Cloud Continuum

“…CI techniques, such as metaheuristics, represent a common solution to explore the search space of NP-Hard problems. Within the resource-management field, these methods have accomplished remarkable results thanks to their capability of efficiently exploring large and complex spaces [35,36]. In particular, CI can provide feasible and robust solutions that are particularly advantageous compared to more-traditional ones [37], which have been effective in some cases, but often struggled with barriers like scalability.…”

“…CI solutions essentially build the knowledge of the target function by collecting the outcomes of their sampling in a memory pool that evolves over time to explore the mostpromising parts of the search space, usually through algorithms that leverage metaphors from the evolutionary and biological world. This scheme has proven very effective in the optimization of static systems, but requires some additional attention in systems with strong dynamical components, as is often the case in the Cloud Continuum [36,41]. In fact, applying CI solutions to solve dynamic and expensive optimization problems opens the problem of which parts (if any) of the memory need to be invalidated and discarded.…”

“…PSO is a simple CI solution that has proven incredibly effective in a wide range of problems, despite requiring careful attention to the parameter setting to facilitate convergence [49]. QPSO is a heavily revised version of PSO inspired by Quantum mechanics, which in our experience has consistently demonstrated solid performance [36,50]. We decided to include PSO and QPSO as relatively simple, robust, and fast-converging CI solutions.…”

“…Then, for QPSO, we configured a swarm of 40 individuals randomly initialized in [0, 15] and a contraction-expansion coefficient α of 0.75. These particular parameter configurations have shown very promising results in different analyses made in the past [36,75], so we decided to keep them also for this one.…”