Since the post mentions that our estimated reads is 10 MB/s and writes is 200 KB/s why bother adding a load balancer. can’t one server handle everything and just have another replica just in case of a failure
Agree. I think that the author just wanted to illustrate or presents a different way to think about the usage of a LB solution (which is implementing high availability). But yes, you are right, in a real world problem like this, a HA cluster (active-passive or active-active) could handle this workload easily.