Scalable Declarative IT System Update Automation by A Search with Critical-Element Heuristics

As IT systems, including network systems using technologies of software defined networking (SDN) and network function virtualisation (NFV), become larger and more complicated, the cost of system management will also increase rapidly. Declarative system update with state space search is a promising approach to enable automation of consistent system update tasks; however, current algorithms is not scalable enough for practical systems. We propose a heuristic algorithm to greatly reduce computation time to solve the system update planning problem for practical systems. Our heuristic algorithm takes into account the structural bottleneck of system update and conducts advance search to resolve bottlenecks of current system states. We evaluated the proposed algorithm for two examples of NFV systems. The results show that computation time to solve system update planning problem for a system with 100 virtual machines does not exceed several minutes, whereas the conventional algorithm is only applicable for a very small system.