I was reading the following about 'sendmail' about the time that I got this
message:

   If the first character of the user name is a vertical bar, the rest of the
   user name is used as the name of a program to pipe the mail to.  It may be
   necessary to quote the name of the user to keep sendmail from suppressing
   the blanks from between arguments.

which prompted me to think of the following possible method of moving your data
to the 9K:
   1. Configure 'sendmail' on the 9K to allow 'mailing' to a program. (I have no
idea how that is done but when I tried using Jim Wochuck's mailer to do that,I
used 'cat', I got the error: ">>554 |cat.. Cannot mail directly to programs". I
hope someone with more 'unix' knowledge can chime in)
   2. Get a hold of a program that will read each record in your dataset and
write it to a message file (have the program write a record with a single '.'
after the last dataset record).
   3. Get a hold of Jim Wochuck's (sp?) mail program, or the one available from
'TELAMON', point it towards your 9K and start mailing the message file. (Use
file equations, i.e. "FILE F1=msgfname,OLD;SHR;LOCK" to write and read from the
message file).
   4. Get a hold of a program for the 9K side that can read from a pipe and
write it to a file or an oracle table on the 9K side.


Another approach maybe:
   1. Go to 'www.telamon.com' and get copies of the 'lzw' utility for the 3K and
the 9K and place them on the corresponding machines.
   2. Extract the dataset to a flat file on the 3K side and use 'lzw' to
compress it (providing you have the disk space).  I've seen very high, 70%/90%,
compression rates for image datasets.
   3. FTP this smaller file to the 9K
   4. Use the 9K version of 'lzw' to expand the file and load it into oracle.


I have used the 'lzw' approach to move 'spoolfiles' from our 3K to a vendor's
SUN machine for creating micro fiche.

Regards

Paul H. Christidis
______________________________ Reply Separator _________________________________
Subject: now what? was tar problem
Author:  James Trudeau <[log in to unmask]> at CCGATE
Date:    12/05/97 15:06

Howdy,

And to all many thanks.

Let me get a bit further into the scenario.  We've got to migrate our
production TurboImage database to an Oracle database on the 9k.
Our third party vendor has provided software (said laughingly) to do
this, but they have apparantly never encounterd a database of this
size.

In this particular instance the program processes through specified
datasets and writes the records out to a flat ASCII file with a record
length of 1024 bytes.  We have some "relatively" large sets one of
which is currently better'n 22 million records of 85 bytes each.  The
extract program generously writes each 85 byte record into a 1024
byte record which is why I equate the output file to a tape file and
let 'er rip.  The test base I am using wrote out a tape file of approx
7GB of which when I tried to reload to FTP or tar or ? to the 9k
I was only able to load 4GB.  Well ok.  It's only a test.  However,
the *real* thing will be 18+GB.

Now complicate this by the fact that I know squat about Posix/Unix.
That's the background for all the recent pleas for help, after all I can
only read so fast.

Anyhow, I'm heading down the pike right now for a cold one and will
be in this weekend to give the dd function a road test.  Anyway you
come at it though it looks like breaking the file down somehow is
also going to be required.  We have DLT drives on the 9k but only
DDS II on the 3k.

I'll let y'all know how it goes and thanks (a *bunch*) for the help.  When
it comes your turn to be cuffed I'll cut ya some slack.

James (gonna get beer) Trudeau
Pits, Texas