You are running a Kubernetes cluster and have deployed several applications on it. One application is particularly sensitive to I/O latency, and you want to ensure it always has access to the fastest storage available on the nodes where it runs. What is the best approach to guarantee this requirement?