The Quantum carousel also moves up and down.
It does? I didn't notice, but I took a video of it in action with the cover off, I'll have another look.
IBM could have had their design be a little thinner if they used metal instead of plastic to make the magazines. It hasn't really caught me as I
leave 1 RU between devices.
Good if you have a full-size rack - mine is 18U and even that's bigger than I intended, I wanted 12U! I really need to reorganise mine and use more of the back face of the rack...
The two flaws I can see in the IBM library are that the screws and nuts holding the magazines together tend to fall apart if an ECO to put blue threadlocker on them hasn't been performed, and I would much prefer the picker lock to be captive rather than going in a storage slot on the back of the library where it can get knocked out and lost. If I get to pick a third one, I'd say that the picker gears should be made out of a much stronger plastic.
Agreed on the picker lock; I was fortunate to get one with the TL2000s and it's clear how stupidly easy it is to lose. Thanks for the PSA about the magazines; I actually have two TL2000s, one is just a pile of parts but had better faceplates than the complete one, so I swapped the fascias and I think there was already threadlocker on the screws. I may just check again and check the TL4000.
And yes, the gears all being plastic does strike me as a real weakness, especially the gear track that the robot moves side to side on. That seems to be asking to strip if the robot gets jammed.
I've used libraries of the "room size" type before and they also have their foibles. But they generally have a repair person on site full-time.
We have two large Spectra TFinity walk-in libraries at work - 20 tape drives each, and enough TS1160 slots to keep a few hundred PB onsite. And yes, those are absolutely mind-bending when they go wrong. We don't have a repair person, but one of the guys on my team manages them and is well versed in their foibles. Unfortunately it's ONLY him, but he's been training me and another guy to troubleshoot in his absence.
The IBM PSU is a unique design but there's a zillion of them for sale out there. Same for library controller cards and the drives themselves (as long as you watch out for OEM lock-in). With the Superloader, it is usually easier/cheaper to buy a whole library with some undesirable tape format and throw the rest away than it is to buy a PSU listed specifically for the Superloader. Hence my article about modifying a standard model.
Yeah, I can see the Quantum library is not designed to be user-serviceable. It is a lot more compact than the IBM design, even if that also means fewer tape slots. Props to IBM for making most components hot-swappable, though that means a LOT of additional empty space inside the unit and a lot more potential for movement. Good that most of the parts are interchangeable between the 2000 and 4000 though, including cards, drives and PSUs. My TL4000 is fully loaded with dual PSUs and doesn't complain when I only run it from one, so I may just run it that way and keep the other PSU in reserve.
Aside from PSUs, I find the magazines prone to lock-up and inconvenient to bulk load (in the IBM, eject the magazine and you get 12 convenient slots - with the Superloader you have to feed them in one at a time through the mail slot). They also don't seem to hold up - I've had 5 of them that have gone to electronics recycling due to drive problems or other issues. I have installed around a dozen of the IBM libraries and they're all still working fine.
Yeah, the paternoster mechanism is clearly complicated. I figure that since the motor only moves them forward and back, and the rest is mechanical, there's less to go wrong in an unfixable way. If the positioning sensors in the IBM go bad, you're stuffed. The Quantum one seems to be simple microswitches, and the most likely mechanical failures are probably fixable by just repositioning the slots, or even with a bit of glue.
And you can still eject the magazines and fill them manually rather than use the mailslot, but I agree its waaaaay more tedious than the IBM approach - you have to move the mechanism to expose the empty slot to the cutout in the casing, rather than having all empty slots ready to use. Took me far longer to fill the 16 slots in the Superloader than to fill the 48 slots in the TL4000!