MEMORANDUM

TO:     Bill Campbell
FROM:   Keith Corbett
DATE:   March 10, 1987
SUBJ:   3.0 Tape Distribution To Customers By Field Service
COPY:   Joe Dreussi, Ron Genest

This memo is to reiterate and clarify my belief that the only real,
viable mechanism for 3.0 tape distribution is to continue to make tapes
in-house; and, further, that we must purchase sufficient new tapes to do
so.

The following suggestion has arisen several times: that we have the
FEs (field engineers) install 3.0 on-site from their own tapes.  The
customers could, in theory, back up the software from disk.

There are several problems with this suggestion [which, by the way, came
up last summer when we first calculated the expense of complete tape
distribution].  These problems are not really unique to the Lambda, and
explain why other vendors do not use field engineers as a primary media
distribution point.

Software is supposed to be installed by the customers.  This is because
a) they have to take responsibility for installation, b) they have to
learn how to do it, and c) having FEs do it is very time-consuming. The
complete 3.0 upgrade for a Plus system takes about 3 hours; someone can
cost this out precisely, but I think it must be prohibitively expensive,
figuring $$'s for a 1/2 day FE time (and extra visits not otherwise
required) vs. perhaps $100 of tape and manual duplication / shipping
costs.  Most importantly, even if we made this a policy, it would take
months for the FEs to cover the many sites (I believe the number is >75)
that don't yet have the new release.

So, this proposed method is

  - more time-consuming for LMI personnel
  - more expensive than normal distribution methods
  - not timely

The idea that customers can make their own backups has several problems.
The boot tapes can only be duplicated from single-user Unix; not al
customers have Unix.  There's no complete documentation on doing the
backups, and customers do this wrong all the time.

Most importantly, from my standpoint, is that rather frequently (at
least weekly), we have to ask some customer to reload software from a
known starting point, i.e. the distribution tapes.  We have to make sure
they have "vanilla" software; we must even confirm the part/rev numbers
on the tapes.  This happens, e.g., when a) they load non-standard or
corrupt software, or b) when hardware problems corrupt the software
(particularly the Unix root and SDU files).  This happens quite
frequently, and is a major headache.

From a legal standpoint, customers with system agreements are entitled,
by contract, to media and documentation from LMI.  The others are
presumably on T&M and should be paying media charges anyway.  If LMI
would/could enforce the T&M rules, we could probably fund the additional
tapes.

I have always favored using FEs to install 3.0 under the following
circumstances:

  1) when the customer is willing to pay T&M rates
  2) for key accounts where no system manager is competent
  3) for sensitive accounts who have been screaming for the update

Whenever we have done this in the past, we have tried to make sure that
immediately following the field installation, the customer receives the
"official" tapes and documentation.  Sending the tapes covers the issues
raised above.  Most important, customers should never receive software
without the accompanying manuals; this generates frustration on their
part, makes LMI look non-professional, and always results in an added
support burden (lots of telephone calls to answer questions that are
covered in the manuals).

The FE distribution scheme does not provide any easy means of tracking
who has what.  By keeping a list in Manufacturing, we can tell who
received 3.0, whether they received the appropriate options (TCP,
Laser1+, Vista/Iris, etc.).  I doubt if each FE could manage the number
of tapes involved, much less could Dispatch track from FAR's what had
been installed.

My final objection to this suggestion is that it is irrelevant; it only
tackles half the problem.  I have been told by Dan Remeika that he is
out of several manuals, and does not have the money to get them printed.
This is a problem that only money will solve.

As for using scrap tapes, I continue to do so as we have agreed, to make
the most out of a bad situation.  The fall-out from bad tapes is higher
on the VAX, but we don't really know whether the scrap tapes cause a
problem in the field.  I can use up my stock of scrap tapes, but there
aren't enough left to do more than about 20 complete sets, out of at
least 75 sets we need to cover each site once (more to cover our
obligation of one set per Lambda).

Sooner or later, no matter what short-term solutions we may find, LMI
must choose whether to purchase materials, or fail to fulfill our
obligations to the customer base.  My feelings are well known, but I
realize we can't spend money we don't have.  I hope this issue will be
resolved soon, one way or the other.  If we are not going to send 3.0,
top management must inform the customers.  At this point many of them
believe we are out of business, and have probably despaired of ever
seeing the new software.

KMC