Two very hopeful products have failed the practical portion of the test.
Mind you, the drives themselves are fine (they’re both 64GB USB 3 drives with “On-the-go” capability so they can be directly connected to a phone or tablet). But the mechanisms for hanging them from my keychain failed.
The first try was the Samsung on the left. The grey thing in the middle is the cap for the micro-USB connector. Note the hole; I used that to loop a cord with a keychain ring on it onto the drive. Worked fine for a few months, until I realized the cord (just nylon, I think, nothing exotic) was cutting through the plastic.
The second was the PNY, which came with the braided metal hanging cable. The mess on the left end of that was a catch that could attach to a keychain loop. But it broke today (that’s the spare piece at bottom left).
I can’t afford to keep buying these things every month or two!
Oracle has dropped Solaris 12 from their roadmap, and while the roadmap shows new SPARC models out to 2020, an acquaintance reports he and the hardware group has been laid off. So that’s kind of the last active remnants of Sun Microsystems, one of the important companies in the history of the field.
It also knocks another processor architecture out of the ring (though I’m sure people will be patching boxes together to keep them running for years). It’s now all Intel, except for phones where the power efficiency is just too important to mess around. (Plus embedded devices, but that’s a different world.)
I see I remembered some details wrong, including the number of platters (it’s 6, not 5).
Quoting the IBM article:
The IBM 1311 Disk Storage Drive provided storage for 2 million characters. Developers of the 1311 engineered twice the recording density of the IBM 1301 Disk Storage Unit by reducing the space between the head and the disk by about a factor of two.
The 1311 used the IBM Disk Pack (later designated the IBM 1316 ), an interchangeable package containing six 14-inch-diameter disks in a four-inch stack, weighing 10 pounds (seen above in the man’s left hand). Each disk surface contained 20 pie-shaped regions. Sectors were segments of track lying within a region, and were the smallest addressable unit, with a capacity of 100 characters. Average access time to any sector was 250 milliseconds, which could be reduced to 150 milliseconds with an optional direct-seek feature. The disks were rotated at 1500 rpm, tracks (50 to the inch) were recorded at up to 1025 bits per inch, and the usual head-to-surface spacing was 125 microinches. The ten recording surfaces provided in normal usage a storage capacity of 2 million characters, the equivalent of approximately 25,000 punched cards or a fifth of a reel of magnetic tape.
We had the “direct-seek” feature, which as I remember it meant the heads didn’t return to the outer edge before starting the next seek. Note those units—some of you are perhaps old enough to remember milliseconds!
Only 1,500 RPM. It’s gradually gone up, so that enterprise high-performance rotating disks today are 15,000 RPM (but lots of people needing that kind of performance are using SSDs instead).
As I recall the 1401 didn’t use the fixed sectoring; or maybe what we used was layered on top of that or something. It’s been a while now. My memories aren’t precise, but we read records considerably bigger than 100 characters as a single operation, and didn’t have to specify the size on each read (I think it was formatted into the pack somehow).
It’s really bad at using system resources effectively. I’m sure this is why it’s so slow at exporting developed photos, and it’s probably also why it’s slow to respond to controls. Here we see Lightroom using barely half the CPU exporting 18 photos—something that is trivially parallelizable to 18 cores (since the photos are independent). In contrast, the old Bibble Pro, which is now available as Corel Aftershot Pro, would always put the whole processor to work.
Recently added monitoring of CPU and disk temperatures to the servers I use. Good timing!
I also replaced a medium-sized fan (maybe 80mm?) with a larger one as the main chassis fan. Since the CPU cooler fan is ducted to the side of the case that probably didn’t affect the CPU temps, but it definitely reduced the hard drive temps.