I have a Kubernetes cluster on IBM Cloud Platform (not important, the question is related to Kubernetes itself).
If I wanted to replicate across different data centers in different regions then, should I use multiple and different master nodes for different regions? What's the best approach in this case and what would you suggest?
Thanks in advance,
I'll answer from an IBM Cloud perspective, since you are referring to data centers.
If you want to "replicate across different data centers in different regions", then you will need to create separate clusters in each of those data centers. Once you have done that, by definition you will have multiple masters (one for each of your clusters). So the short answer is yes, you will have multiple clusters (and masters).
See this doc for more info. In this case you're talking about scenario 3: https://console.bluemix.net/docs/containers/cs_clusters.html#planning_clusters
Note that you will need to provision a global load balancer to load balance between regions, as well as ensure your app can handle any data replication between regions that is needed.