This is a weird one (I think).

We have a requirement to store some information in an encrypted ASCII file 
(that is, it was ASCII prior to being encrypted) on a distributed platform over 
which we have no control.  We also have a requirement that we make sure that no 
data is lost during transmission.

Currently we are creating this as an EBCDIC z/OS PS file.  We also have a 
process that reads in the file and creates an MD5 hash value based on it.  Both 
the data file and the MD5 hash "file" are transmitted using FTP binary mode to 
a distributed FTP server.  A process is run to recalculate the MD5 hash and 
make sure they match.

The idea at this point is that a process run on the distributed side will 
convert the file from EBCDIC to ASCII, adding ASCII line delimiters (based on 
outside knowledge regarding the fixed length of the records in the file).  This 
would then be encrypted and then transmitted to the final destination (in 
binary mode, I hope!!).

That's had enough.  What's worse is that we need to be able to prove we can 
receive the file back on the mainframe and 'import' it back to the original 
data structure.  The thought (not mine!) at this point is that the distributed 
platform will 1) retrieve the file; 2) decrypt the file, and 3) send it back to 
the mainframe using FTP binary mode.  Of course that means it's now in ASCII, 
with ASCII delimiters, on the mainframe.  We then need to convert it back to 
EBCDIC so that we can load it back to the source data structures.

I have not yet tried anything.  One thing I am wondering is if it might make 
sense to place the ASCII file in a z/OS Unix file that is defined as being 
ASCII.  If I copy that to an MVS PS dataset will it convert automatically from 
ASCII to EBCDIC?  Maybe not.  If I copy it from a Unix "ASCII" file to a Unix 
"EBCDIC" file will it convert it?

The problem with either one of these, assuming they'd even work, is that we 
currently have NO use of z/OS Unix for our business applications.  Our systems 
group uses it when they have no alternative, but not beyond that as far as I am 
aware.

If we do go forward using the Unix file system it seems like we probably can 
greatly simplify the process by actually writing the original file directly to 
it a Unix ASCII file with appropriate delimiters.  We could then do the MD5 has 
on that and send it to our distributed FTP server in binary mode.  All they 
would have to do is encrypt it and send it on.

Are there any (easy?) options I should consider that don't use Unix files?

Thanks, Frank


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to