Backups are every good Sysadmin's 'favorite' recurring task.

A good backup infrastructure can mean many things depending on the scale, size, and criticality of data.

In the case of my lab, I treat it as a production-grade hypervisor with RTO of within 24 hours and RPO of within 24 hours. This means that I run my backups once daily and find it acceptable to lose up to 24 hours of data between backups (Recovery Point Objective) as well as that should I need to recover from a failure, I can restore within 24 hours(Recovery Time Objective.) In most environments the RTO would be significantly lower but I run on a very limited budget.

As this is treated as production-grade it needs a proper 3-2-1 backup strategy. For this I use both TrueNas and Proxmox Backups Server (PBS.)

VMs are backed up through Proxmox Backups Server, to keep backup speeds high the backups server keeps it's Synchronous Log file (SLOG) on SSDs and a pair of NVME drives to be a redundant 'Special Vdev' that keeps metadata.

This is replicated to an offsite Proxmox Backups Server via VPN tunnel & PBS replication.

This gives me two of my three copies of data, one off-site.

My third copy of data is a replication of the production TrueNAS storage appliance to a secondary TrueNAS server.

If budget allowed I would also add an LTO tape rotation at this scale.