Rotating magnetic memory is one of the oldest data-storage
technologies.
Drums solve the flying-head problem; you can bolt the heads in
place. Some are contact, but many have a fixed gap.
Drums have lower latency, 1 revolution worst-case. Hence they were
used as main memory for modest machines and "backing store"
(2nd-level) for many 1st-gen machines.
Didn't EDSAC have a drum backing store added? There was one in the
ACE design. Late 1940's saw magnetic drums in use.
I'm too lazy to go out to the lab to look [in books] but
certainly, IBM did not invent the disk drive. They may have
commercialized it first, but I doubt that even. Maybe they were
the first "successful" but that's a stupid measure. BFD. Hand-wave
-- it was obvious.
Radial or axial -- sigh, this issue comes up all the time. Not
axial or radial heads, but the myopia of 'today is the measure of
yesterday'. There were plenty of fixed-head radial disks, like
flat drums.
Today, memory and disk "we all know" what they mean (memory=CPU
local, disk=one-or-more-level-removed) , but that arrangement is
not obvious and has only ruled for little more than half the time
that computers have existed.
1963/earlier:
"hundred million bit disk file... each housed in a 2-1/2 ton
utility truck"
http://wps.com/projects/ARTOC/index.html