To avoid running into that situation again you normally just create many LVs out of the one VG on your block device.
There's no limitation like that, and it works well.
If you want to do things like having a VG for the OS and one for data, then you would need to partition the block device (i.e. i have a 16GB partition for the OS -> vg00 and the rest is another partition for all data volumes -> vg01)
But this is not required, you can do fine with one VG.
For the current issue:
- Not recommended
In Linux, it is possible to create a PV on an LV, so you could add another VG on the existing space fully online. But it would be pretty ugly and probably fall on your head a year later.
- Not recommended
The other option is to shrink the existing VG using the pvresize command.
(and adjusting partitions. This is more risky. I'd recommend to be really careful and test in a VM first)
Once the first PV (and thus the VG) has been shrunk you could repartition the disk and create a new partition, that will be part another VG.
This doesn't even work if the existing PV has data spread out to the end with gaps.
I feel you can simply work with more LVs and forget about the new volume group, maybe you just didn't yet know that there's no limit on how many LVs you use?
one hack I've actually used (required test-runs and ended with an hour of downtime) was to add another disk via iSCSI, mirror to that, and then rebuild my "local" storage. The problem is you need to boot off a rescue media to shrink your root partition (shrink is offline for anything but veritas vxfs).