Sony sees the cold light of optical archives, buys ex-Facebooker's upstart Sony is buying ex-FaceBook open compute project exec Frank Frankovsky’s optical storage startup to boost its optical archive prospects. Frankovsky’s idea was to use racks of enhanced-capacity optical storage disks for archiving cold data. The system would have better throughput than existing optical storage systems. He started …
It's only because of a misunderstanding of how tape is written -- tape inventory is read on every mount and written to every time there was a write to tape, so this region of tape sees heavy use and without the inventory, the data may be safe, but seek times will be dreadful.
Hence cartridge memory in an RFID chip, but it's not a magic bullet, since EEPROM also has its limits.
Quite more than a few times, in fact. That alone probably kills the idea of an optical archive. Well, that, and the fact that a lot of people had very bad experience with optical media.
There are two small high-density rackmounted libraries on the market. BDT makes one for a number of OEMs, and Oracle makes SL150.
BDT's density is 8 cartridges in 1 rack unit, and then 24 cartridges per 2 rack units in 2, 4 and 8 unit form factors (top density of 12 carts/RU).
SL150 scales from 30 cartridges in 3 rack units to 300 in 21 units (top density of 14+ carts/RU).
At LTO-6 cartridge density, this adds up to 30-35.7 TB/RU.
This Sony contraption looks like 4 rack unit height, so it's only 12.5 TB/RU. And mind you, those high density bluray disks are not cheaper than tape per byte.
If as the article suggests the plan is for cold data, then in many situations the data already archived will never change and thus the rewriteability is not much of a problem. I would be more concerned with the classical issue of optical disks for backup/archive and that is a to high price/GB ratio compared to possible advantages to make much sense.
Rather than have to retrieve a disk and return it to a drive, it would make sense to have a drive shaft and optics on the arm that would otherwise retrieve the disk in the first place.
There was one suitably daft idea that I laughed out of my head ages ago, which was rather than rotate the disk, why not rotate the laser. But I thought that I must be nuts, so chastised myself for being stupid; the pits would have to be burned in such a way that the reflection of the laser always went back to a pre-defined point ... ok, not beyond do-ability but it would then make more sense to completely do away with the disk format and go for something else ... at which point my brain went, "rum and coke..." and never progressed beyond that point.
I've been stuck in a bottle ever since...
"Rather than have to retrieve a disk and return it to a drive, it would make sense to have a drive shaft and optics on the arm that would otherwise retrieve the disk in the first place."
I was thinking the same thing. The one immediate drawback of this approach I could see was that the arm mechanism would be much larger than a simple grappling arm (to retrieve/return discs) so the discs would have to be spaced much further apart, thus limiting the number you could cram in the box.
Ummm.
circa 70's era juke boxes did this.
with DVD/BlueRay to get the data throughput you'd want putting the optics in the retrieval arm means that you have your 1 dedicated arm tied up while the disk is read off. The model presented allows you to stuff the disk in the drive and go get other disks to satisfy io requests.
other point, air flow becomes an issue if you have the optics on the moving bit, it will alter the head flight.
This has a place for WORM data. With allowed higher latency. That you don't call back all that often.
It's an interesting thought, yes, but the capacities are just too small to be practical at this point. BluRay has fallen behind the times which is why I'm eyeing Archival Disc with some anticipation. Not only are the disc sizes closer to what's needed for large consumer backups in the terabyte range, but they're designed for "cold" storage.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
