What most folks take for granted today even more than the size of drives is the reliablity. When I graduated in '84, I went to work for Harris RF in Rochester, NY. There, they had a PDP-11/70 that had not one, but two (!) 300MB RM05 drives, the DEC equivalent of the CDC 9766 300MB drive. I thought I was in hacker heaven. They were the size of washing machines and you had to mount the 12-platter disk pack into the machine from the top. If you didn't take special care in the handling of the disk pack, you'd have a head crash.

Here's a picture of a DEC RM05 disk pack with a semi-modern 20GB disk drive sitting on top:

http://www.psych.usyd.edu.au/pdp-11/Images/rm05.jpeg

That disk pack is in the storage container. To mount it in the drive, you'd pull off the bottom of the container -- which was much like the bottom of a cake container. There was a "T" style handle on the top of that plastic bell you would use to pick up the disk pack, all 12 platters. You'd open the top of the disk drive (which was literally the size of a top-loading washing machine), you'd place the disk pack over the threaded spindle in the drive, you'd screw down the platter stack onto the spindle, which would release the platter stack from the plastic container, you'd pull up on the T-handle and come away with the plastic bell. Now you'd close the disk drive, get a status indication on the front panel that the drive was closed. Then you'd press a button on the front panel and spin the drive up to speed. You'd see the button glow red, and would hear the motor spinning the disk platter mass up to speed. When speed was achieved, you'd see the button go from red to white, meaning the drive was ready for you to load the heads. You'd then hit a second button on the front panel to load the heads onto the platters. If that went well, you'd get another white light indication, which meant that the disk was ready for use.

When something went wrong, you lost data. Fast.

"Special care" took in a wide variety of issues -- you needed to be scrupulously clean. You could not smoke anywhere near these disk packs or the drives. They needed filtered air. You couldn't bang the disk packs or the drive around. If you dropped the disk pack, it was ruined. If you got the disk pack near a magnet, motor, transformer, etc, you lost data.

We had two CDC 9766 drives on another minicomputer in the machine room in 1984 -- a Harris H-800. (What a piece of crap the H-series minicomputers were... but that's another story). One winter Saturday evening, a Harris computer tech came up from Melbourne, FL, to diagnose some problem with the H-800's backplane. The guy flew into the Rochester airport. Back then, computer techs had hard drive disk packs, tapes and paper tapes (remember loading programs off punched paper tape?) of diagnostic programs. This tech had rented a car at the airport and schlepped all his stuff over from the airport. He walzes into the machine room, brings down the H-800, dismounts the system disk pack, mounts up his diagnostic pack, spins it up and tries to reboot the H-800. This is about 1500 in the afternoon.

H-800 doesn't boot his diagnostic pack.

Anyone who has used these old hard drives in the past has probably already rolled their eyes and knows exactly where this is going. But most recent computer users who have never used a removable disk drive from the 70's and early 80's are not yet clueful about what has happened, so I'll continue telling the story.

The tech fiddles around with the situation for a while, then decides to go get dinner. He leaves his diagnostic pack loaded in the system drive (equivalent to the "C:" drive on a Windows machine). It is now about 1800.

I was working there and at about 1900, I go into the machine room to mount a tape on the PDP-11/70. I notice the system disk drive for the H-800 ain't sounding real happy, but I noticed the tech's toolcase and stuff laying there, so I figured the Harris tech had things under control. I continue doing what I'm doing on the PDP-11/70 that was arm's length away from the H-800 and get back to work. At about 2100, I go back in to dismount my tape and I notice that the air in the machine room is brown. The air in the room is literally a brown haze. I also notice that the Harris tech is nowhere to be seen, his tools are gone and the H-800's front panel lights indicate that the machine has power, but isn't running, ie, isn't "booted." I had a quick look around and get the heck out of there, 'cuz even at my tender and naive age of 22, I've got this feeling in my gut that there's some serious turdage about to hit the turbine blades.

Fast-foward to Monday morning. Man oh man, there is a near riot in the hallways. The system admin for the H-800 is stalking the hallways, looking like he's going to kill someone. Everyone in manufacturing, radio hardware design, inventory, PCB routing, etc, etc is wandering in the hallways. 30 minutes goes by -- it is 0830. Word goes out that there is to be a meeting for all H-800 users (who numbered about 300) in the company cafeteria at 0900. At about 0845, I am called into my director's office with the H-800 admin. I'm quizzed as to what I know because the security logs showed I was in the computer room over the weekend, and I relate what I know and the timeline.

Next thing I know, my director, the H-800 admin (also a director, but of hardware), my supervisor and I are all on the elevator up to the VP of engineering's office, where we pick up the VP and are on our way to the CEO's office. I'm ordered to relate my observations to the President/CEO, whereupon I'm thanked for my time and told to return to my cubical to do what I'm supposed to do, which was write s/w on a VAX for PDP-11's, having nothing to do with the H-800.

I was told an hour later that after I left the CEO's office that there was some "serious yelling" that occurred in the CEO's office, then a phone call was made to the CEO of Harris' minicomputer division in Florida, and there was MUCH more yelling done at some "seriously elevated volume levels."

Here's what had happened:

The tech from Florida came into Rochester when it was about 8 degrees F above zero outside. He was a native Floridian, and knew nothing about this thing called "winter," much less anything about this idea of "cold" and what happened as a result of cold. One of the Very Important Rules of those old removable hard disk drives was that you had to allow the disk packs to achieve ambient machine room temperature before you loaded the heads onto the platters -- you were supposed to put the disk pack into the drive, but NOT hit the "load" button to load the heads for, oh, two hours. If you jumped the gun and loaded the heads early, there was a near-certain possibility of a disk->head crash when you pushed the button to load the heads.

Now, it needs to be explained that there were all sorts of "head crashes" in those days. A "minor" head crash would simply corrupt the media -- ie, ruin the disk pack. If you looked very closely, you'd see a very minute scoring in the surface of the disk. A serious head crash would scratch the heads so badly that you could see it immediately -- this level of damage would crash the heads on the drive, which would then ruin any other disk pack you'd subsequently mount in the same drive. The clearance between the head and the disk was such that a cigarette smoke particle wouldn't fit between the disk head and the disk -- that gives you an idea of the tolerances here.

Here's a picture of disk heads from those days:

http://www.psych.usyd.edu.au/pdp-11/Images/head_size.jpeg

The PC disk is the same 20GB disk from the above picture. You get the idea here -- disks and their hardware were big in those days.

That Monday morning, the H-800 admin found the disk heads were basically gone, completely burned off the end of the mechanical assembly that held the heads. They had scraped most of the ferric oxide off the disk surface, leaving nothing but the aluminum platter -- and this was the reason why the air was brown in the machine room.

Now, it was not enough for the Harris computer tech to: a) crash his own disk pack, but because he left the pack in there, he crashed the heads, and when be b) remounted the original system pack, that disk pack was destroyed.

But wait! He wasn't done yet. He mounted his crashed disk pack (which was by now at ambient temp, but very badly ruined) into the "user data" (the equivalent of a Windows "D:" drive) disk drive, crashed those heads, and then remounted the user data disk pack into the drive, thereby destroying the 300MB user data disk. Both disk drives had their heads destroyed, both disk packs were destroyed, all the data was lost. It cost the company three days to get the drives repaired, to get new disk packs, and it costs a week more to get all the data put back where it belonged, even with weekly complete and daily incremental backups.

Oh, and it cost the tech his job, and we billed the computer division for the lost time, hardware and media.

That's what disk drives used to be like. I don't miss 'em a bit. ;-)