About a month ago, I posted a request:
>I'm looking for a device that will feed an Exabyte drive with a series
>of tapes in sequence (so I don't have to come in to work at 3AM when
>the drive runs out of data). I only need sequential feeding, not
>I've found two devices that seem suitable. The EXB-10 is a 10-tape
>robotic-arm mechanism made by Exabyte. The mechanism is capable of
>random access, and there is a "10i" version that provides random
>access, but the "10" is a cheaper sequential-access version.
>The other is the ACL-600, from a company called Automated Cartridge
>Libraries. Its mechanism is sequential-only: the input tapes just sit
>in a gravity-feed stack, and tapes are dropped into an output bin after
>being read. Whenever the drive door opens, the mechanism extracts the
>tape, drops it in the out bin, allows another tape to gravity-feed from
>the input bin, and then closes the door. It's something like half the
>price of the EXB-10.
>What I really care about is reliability. The ACL-600 is mechanically
>simpler, but depends on gravity for some things - it could end up being
>less reliable. If anyone has real-world experience using either one
>of these devices, please send me mail about your impressions (good AND bad).
Well, a bunch of people commented on the reliability of gravity. What I
*meant* was that the gravity-feed device depends on the walls of chutes
and bins passively controlling the motion of the tape, and it's at
least conceivable that such arrangements would jam occasionally. A robot
that actively grasps the cartridge and moves it under power has the
potential for more precisely-controlled cartridge motion, plus a better
ability to detect jams when they happen. I really wasn't concerned with
gravity itself failing.
I received a few reports from present owners of both the ACL and Exabyte
devices. All were positive - nobody had any problems with either. I
also talked to three dealers. The one dealer (who sells both) said that
a higher percentage of the ACL devices had problems that required warranty
repair than did the Exabyte devices. However, further questioning
suggested that these were mostly drive alignment problems, and that once
an ACL was set up properly it also worked reliably. The fact that the
ACL's drive is normally installed by the user while the Exabyte's is
normally installed by the factory probably has some influence on this.
The other two dealers, who each sold only one of the devices, thought
that whichever one they sold was better.
I was quoted prices of $1700-2000 US for the ACL unit. The Exabyte
10i (random-access) was about $3800 US or $4540 CDN, plus extra
for software to control it. The sequential-only EXB 10 was $3490 CDN.
Advantages of the Exabyte unit: a random access version is possible.
It holds 10 tapes. It's a neater package. It's probably gentler in
its tape handling.
Advantages of the ACL: it's half the price. It's mechanically simpler -
probably less to fail. It's pure sequential, with separate "in" and "out"
bins. If you are reading or writing more tapes than will fit in the
device at one time, reloading the Exabyte requires that the operator
figure out which tapes have already been read and which ones have yet
to be read, remove the former only, and then fill the slots with the
next new tapes in the correct sequence. With the ACL stacker, the
operator just empties the "out" bin, and stacks more tapes in the "in"
bin until it is full. There's only one place to put tapes, so you
can't put them in the wrong place.
I bought an ACL stacker, and have had it assembled and running for
a few days. It was not hard to install the drive in it; ACL supplies
instructions and an alignment gauge. It has worked flawlessly so far.
It came with front feet that are quite a bit higher than the rear feet,
tilting the whole unit backwards a few degrees - I assume that someone
found that this increased the reliability of either loading or unloading
tapes. It does look a bit weird tilted backwards like that.
I did find that the two optical sensors that look at the LEDs on the
drive was not perfectly aligned with the LEDs themselves, and there was
some crosstalk. This is probably because the 8200 uses vertical LEDs
and the 8500 uses horizontal LEDs, so the position of the sensors is
necessarily a compromise. I found that bending the the "arm" carrying
the optical sensors downwards a few mm improved things. Other than that,
the stacker did not need any adjustments.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:51 CDT