Drum HDDs?

tServo · 2020-08-05, 16:31

Just wondering:
Since access speed is one of the 2 most often mentioned reasons for switching to SSDs, why doesn't somebody offer a HDD with a r/w head that doesn't have to move?
IE, the arm would be in a fixed position and it would have a r/w "head' for every track.
I remember back in the 60s working on some mainframes ( CDC ) that had 1 or 2 drums for the operating system files and ( much ) slower disks for everything else.
Any ideas, Laurv ?

chalsall · 2020-08-05, 16:34

Quote:

Originally Posted by tServo

IE, the arm would be in a fixed position and it would have a r/w "head' for every track.

Do you have any idea how thin the tracks are? Then determine the smallest possible R/W head.

Spinning polished rust only makes sense for long-term storage nowadays.

Uncwilly · 2020-08-05, 16:48

Weren't the drum or cylinder drives single head and designed for more sequential access?
And at one point I heard that there were machines that pulled the cylinders out of an array to read them.

retina · 2020-08-05, 22:32

Quote:

Originally Posted by chalsall

Spinning polished rust only makes sense for long-term storage nowadays.

Nonsense. My entire lair has no SSDs, they are too small for the price (and really hard to ensure erasure).

chalsall · 2020-08-05, 22:41

Quote:

Originally Posted by retina

(and really hard to ensure erasure).

Surely ("Don't call me Shirley") you have a fireplace (or a crucible) in your lair?

kriesel · 2020-08-05, 23:33

There are at least four determinants of rotating storage data rate;
rotational speed,
head seek time,
linear bit density,
interface data rate.
Caching on the drive assembly or upstream reduces the impact of some of those.
Some early physically large drives used many heads per surface and very short head motion.

Durability is another consideration. https://superuser.com/questions/7899...roof-is-an-ssd

chalsall · 2020-08-05, 23:41

Quote:

Originally Posted by kriesel

Some early physically large drives used many heads per surface and very short head motion.

Some early compute used soundwaves in tubes of mercury for short-term storage.

Shouldn't we, in 2020, be past spinning stuff?

Heck, even our gyroscopes are now solid-state.

Chuck · 2020-08-06, 01:01

Quote:

Originally Posted by tServo

Just wondering:
Since access speed is one of the 2 most often mentioned reasons for switching to SSDs, why doesn't somebody offer a HDD with a r/w head that doesn't have to move?
IE, the arm would be in a fixed position and it would have a r/w "head' for every track.
I remember back in the 60s working on some mainframes ( CDC ) that had 1 or 2 drums for the operating system files and ( much ) slower disks for everything else.
Any ideas, Laurv ?

The UNIVAC FH-432 (faster but lower capacity) and FH-880 (slower, higher capacity) were the two head-per-track subsystems. FH=flying head

Aramis Wyler · 2020-08-06, 01:24

Quote:

Originally Posted by tServo

Just wondering:
Since access speed is one of the 2 most often mentioned reasons for switching to SSDs, why doesn't somebody offer a HDD with a r/w head that doesn't have to move?

You could do that, but it would be cheaper, faster, and more effective to take commercially available drives and put them in a RAID 0 or 5 configuration, depending on the number of drives. Since the drive arrays in those RAID configs are treated like a single drive, you gain the advantage of 5 - 6 arms with 16-25 heads writing to both sides of 15-24 platters as a single action into a 48-192MB cache on regular commercial motherboards.

But would still not as cool, quiet, or fast as a solid state disk, though it would remain re-writeable for longer if you could keep replacing the drives as they fail and until recently would be a good bit cheaper.

My main machine, which has been down for a while but I which I am resurrecting tomorrow, has 4 slow, quiet, low energy 512GB Seagates raided up into a single drive. It was both faster and quieter than most Raptors. These days though, I'd just cough up the $150 for a mid-range 1TB SSD.

retina · 2020-08-06, 02:58

Quote:

Originally Posted by chalsall

Some early compute used soundwaves in tubes of mercury for short-term storage.

Shouldn't we, in 2020, be past spinning stuff?

Why? Just because something is old doesn't mean it is bad. Just because it is new doesn't mean it is good.

My sensitive point is price-per-bit. So HDDs win.

Uncwilly · 2020-08-06, 03:10

Quote:

Originally Posted by retina

My sensitive point is price-per-bit. So HDDs win.

Buy a 25 kg bag of sand and use each grain as a bit. That is cheap.

2020-08-05, 16:31	#1
tServo "Marv" May 2009 near the Tannhäuser Gate 3·269 Posts	Drum HDDs? Just wondering: Since access speed is one of the 2 most often mentioned reasons for switching to SSDs, why doesn't somebody offer a HDD with a r/w head that doesn't have to move? IE, the arm would be in a fixed position and it would have a r/w "head' for every track. I remember back in the 60s working on some mainframes ( CDC ) that had 1 or 2 drums for the operating system files and ( much ) slower disks for everything else. Any ideas, Laurv ?

2020-08-05, 23:33	#6
kriesel "TF79LL86GIMPS96gpu17" Mar 2017 US midwest 2⁴·3·163 Posts	There are at least four determinants of rotating storage data rate; rotational speed, head seek time, linear bit density, interface data rate. Caching on the drive assembly or upstream reduces the impact of some of those. Some early physically large drives used many heads per surface and very short head motion. Durability is another consideration. https://superuser.com/questions/7899...roof-is-an-ssd Last fiddled with by kriesel on 2020-08-05 at 23:37

2020-08-05, 16:48	#3
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 2²×2,767 Posts	Weren't the drum or cylinder drives single head and designed for more sequential access? And at one point I heard that there were machines that pulled the cylinders out of an array to read them.