VMware Cloud on AWS ESXi hosts deploy with a ~4096 FD limit. This is a tuneable ESXi parameter; given how *large* the hosts are, it makes sense for this to be higher. In a Linux environment, an extra 4k FDs only "costs" about 1MB of RAM, and given that VMC has >500,000 MB of ram per host, this seems like a small price to pay - and would only be consumed when needed, most likely.
Additionally, neither vCenter nor SRM "enforce" any practical limits (via limiting # of disks, # of replications/replication points, etc), so this is a "bound" that a customer can easily cross without knowing it.
|Area related||SDDC scale|
|Other impact on customer||FD exhaustion can have catastrophic risks from side effects|