[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
FORCED to buy new cart for XL400?
Date: Mon, 7 Aug 89 09:32 EDT
From: JR@YUKON.SCRC.Symbolics.COM (Johanna Rothman)
As one of the people who made the decisions about the QIC-100 vs. QIC-11
tape drives and software distribution, I will be happy to answer the
specific questions below, and the philosophy behind the issues of
QIC-100 vs. QIC-11.
The general philosophy is that if we come out with an embedded product,
the software distribution mechanism must match the mechanism of the
host. When we decided to implement MacIvory, we decided that the
"peaceful co-existence" route was the right way to go. We also decided
the MacIvory had to be able to be a stand-alone machine, so that it
really could be used as a delivery vehicle. This means that we had to
use tape drives compatible with the Macintosh, not tape drives
compatible with the 36xx-series machines.
Is QIC-100 the only tape drive compatible with the MacIntosh? If not, then
this argument doesn't support the selection of QIC-100. (I myself am not
sufficiently experienced with MacII's out in the real world to be able to
answer my question.)
When it came time to decide what tape drive to use on the XL400, we
decided that since the Ivory software was the same, we would use the
same tape drive, and we wouldn't have to cross master and installation
By the same argument of compatibility you use above, it seems to me you should have
chosen QIC-11. The installation testing is a non-issue, as far as I'm
concerned, see below. Many customers have requested support for QIC-11 on
ivory-based products, and Nihon Symbolics went so far as to pay consulting to
produce the software to do this for one collection of hardware (Emulex controller
+ any QIC-02 drive).
It's not conceptually difficult, or even difficult in practice to write
the contents of one kind of tape onto another kind of tape. The
mastering is a very small part of the tape generation process. The
larger part of the tape generation process is installation-testing and
The SQA group spends its entire resources installation testing a release
for at least 4 weeks, to master and test all the tapes for a release.
(It generally takes closer to 8 weeks for all products.) If we were to
cross-master, then we would spend at least half those 4 weeks on every
This seems like a specious argument to me; what you are testing is the CONTENTS
of the tapes, i.e. that the set of files on the tapes will properly install.
If you have the same set of files on several different media, you only have to
do this type of testing once, not once per medium. You do need to verify that
the contents of the bits on tapes of all the different media are the same, but
this should be an automatic part of the tape duplication process, nothing that
requires human resources.
Then we would have to be able to duplicate the software
onto the appropriate media, and we would have to ship every customer a
custom kit of upgrade software.
I don't think the customer was asking for this; just that the set of files
distributed as a release be available on both the new and old media, so customers
with many of the old drives don't have to buy a new one. This seems like an
eminently reasonable request to me.
Right now, we are able to have a software update kits packed for us by a
vendor. If we were to allow our customers to order software updates by
the kind of tape you wanted, we would have to kit and pack the software
by hand ourselves. That would add an enormous cost.
Yes, but that wasn't what the customer wanted (as I read his mail at any rate).
The problem we have is this: There are many customers who already have
36xx machines with cart tape (QIC-11). Some of those people have bought
MacIvories or XL400s. We have some customers who have already bought
MacIvories or XL400s with cart tape drives (QIC-100) who are now buying
refurbished 36xx machines. We cannot afford the time to cross-master,
installation test, and duplicate every tape we produce.
You yourself said that the mastering was a small part of the process; as far as I
can see, that's the only additional work you have to do. The number of duplications
is essentially the same, especially if you register each customer as requiring a
particular type. Then instead of making n+m of one type, you make n of one and
m of the other, for a total that is the same. (Yes, I realize you'd have to keep
a few extras for spares inventory, but I don't think those few extra copies would
invalidate my argument).
DEC successfully used to ship its software on a lot more media than we'd ever have.