I'm buying hard disks for backup imaging, which (assuming the software and controllers are working correctly) is essentially a single long sequential read from one and write to the other. I want these operations to run as fast as possible.
Does spindle RPM constrain the speed of such sequential operations at any point (e.g., near the inner edge of the platters where the drive head is covering disk area at the slowest rate)?
If not, what is the limiting factor in data transfer speed between two spinning disks?
The limiting factor is the rate at which data can be transferred to and from the platters. The bits-per-inch value is the same for each zone, so the innermost zones have the least bits-per-track. Less bits-per-track means less bits-per-second.
For a typical 3.5" drive, the ratio between the outermost and innermost transfer rates is about 2:1.
So... it sounds like you're saying that the practical I/O limit to a drive is constrained by the linear speed of the R/W head, and that speed IS proportional to the spindle RPM. But in practice is that the limit on the drive's I/O? E.g., will the observed sequential I/O on a 3.5" 5400rpm drive be limited by its platter speed, and if so does that apply across the platters or only on the inner zones?
You can confirm the relationship between RPM and transfer rate by comparing a 5400 RPM drive against a 7200 RPM drive from the same generation. Assuming that their platter densities are identical (eg 1TB per platter), and assuming that their sector sizes are the same (either both are Advanced Format models or both are non-AF), then their transfer rates will be proportional to their RPM. If you compare a non-AF drive against an AF model, then the transfer rate will be about 10% faster for the latter. That's because its 4KB physical sectors pack about 10% more data bits per track that the equivalent 512-byte sectored model.