Bare with me -- New user playing with the trial version & almost ready to buy it but have a few questions. My setup is 7 VM's on ESXi, vSphere Client, hosted in a data center on my own Dell R710 box. Speedtest gives 100Mbit/sec both ways. Backup storage is in an EC2 bucket.

The ideal setup would be to run the backup program from an EC2 instance with the machine on 24/7. However, full backups take way too long. An hour for a 10gig provisioned machine (and that's the smallest of my VM's) with less than 1 gig of actual storage for the backup.

The above lead me to believe reading from my server was the bottleneck, so I installed a copy on one of the windows server VM's (not the one being backed up) and it only took 10 minutes to do a full backup.

Restores are a different story. Running from EC2 is naturally quicker, since it's reading from S3.

Naturally the other issue is even if I were to accept the slow restore speed from ESXi since that will hopefully be never and reading/backup is the more frequent operation, if a hardware failure were to occur and the whole system needed to be rebuilt, the program would have to be run either in EC2 or on another computer to rebuild a new server.

I realize I can call support to release a "stuck" license in the event of a disaster, but is there any reason why Cloudberry couldn't issue 2 licenses (and if you really wanted to be strict about it, make sure the second can only access the repository of the first) so we can get optimal performance on both sides and not have to worry about one more issue in the event of needing to recover from a disaster?

Second question(s) -- maybe I'm doing something wrong in the planning stage. Is EC2 even the right place to host? Are there default settings I should change or something else I should to do to increase read performance from my server? FYI -- I've tested a t2.small and an m5.large EC2 instance & it did not make much of a difference. What do most of you do in a similiar setup?

Thanks,
I want to get this correct before implementing a permanent plan and backing up nearly a TB's worth of VM's.