To determine the Pod per node CIDR range, you need to calculate how many IP addresses are required for each node, and then choose the smallest CIDR range that can accommodate that number. A CIDR range of /n means that there are 2^(32-n) IP addresses available in that range. For example, a /24 range has 2^(32-24) = 256 IP addresses.
According to the question, the application team requires a minimum of 60 Pods per node and a maximum of 100 Pods per node. Therefore, you need to choose a CIDR range that can provide at least 100 IP addresses per node, but not more than necessary. A /25 range has 2^(32-25) = 128 IP addresses, which is enough for 100 Pods per node. A /26 range has 2^(32-26) = 64 IP addresses, which is not enough for 60 Pods per node. A /24 range has 256 IP addresses, which is more than needed and wastes IP address space. A /28 range has 2^(32-28) = 16 IP addresses, which is far too small for any node.
Therefore, the best option is B. /25. This is also consistent with the Google Kubernetes Engine documentation, which states that each node is allocated a /24 range of IP addresses for Pods by default, but the maximum number of Pods per node is 1101. This means that there are approximately twice as many available IP addresses as possible Pods, which is similar to the ratio of 128 to 100 in the /25 range.
1: Configure maximum Pods per node | Google Kubernetes Engine (GKE) | Google Cloud