We present solutions for reliable mulit-resource management for network infrastructures. We propose enhancements to four distributed multi-resource allocation policies for a network slice provisioning for robustness against resource controllers failures. We define two cascading and two parallel computation approaches; their comparison, using a simulation approach, show that the parallel ones perform better in terms of fault resiliency and provisioning latency, while a cascading approach using a controller pre-ordering policy is able to better reallocature resources by recomputation upon failures.