Alexander, could you please clarify the following: • The same recommendation stands for kubernetes deployments right? • Going back to the docs, I see that it seems to recommend i3 instances for bookies and brokers. And t2 for zookeeper. Would a t2 or other general purpose compute should suffice for other pods (zookeeper, toolset, recovery, pulsar-manager, proxy, prometheus, grafana)?

What are my options when I am shifting away from EBS? Tiered storage? Replication? I am worried that instance memory is ephemeral that I can lose data even if there are hardware faults in AWS. I am very new to bookkeepers and I am still learning to manage then and my nightmares are mostly about bookies facing an error like data loss or corruption that they cannot recover from 😅