I have a fair sized installed with 2700 devices and 65k ports. I have also hitting high RRD disks io usage. I finally fixed the problem by moving the librenms VM and its data disks to nvme only storage.
You could try to give the VM more vRAM, so the filecaching can mitigate the load to some extent. Also, activating RRDCached could help a lot.
I also 2nd the hint from @Zach_Underwood. Running on fast storage is always a good idea for tools that permanently write random data.
If your hypervisor has lots or spare RAM, you could also create some RAM-disk and mount the folder /opt/librenms/rrd there.