RE: Separation of Duties...

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

> From: John Myers
>
> No one likes bureaucracy, but the purpose of separation of duties is
> clear ... that it requires a conspiracy of at least two people in
> order for fraud to occur.  For example, a "trusted" person to change
> inventory quantities in a file (ENDJRN, DFU/DBU, STRJRN anyone) & a
> person in the warehouse to remove the inventory.  How many of your
> systems could catch this?

Okay, I'm not going to get involved in a debate, but every system I've ever
implemented since the 1980's could handle this, precisely because the ENDJRN
is posted in the journal.  There is never a good reason to end journaling on
a journaled production file, and if you understand the system, ENDJRN raises
an immediate flag and an investigation.  If the "trusted" person who ran the
ENDJRN doesn't have a good explanation, they probably should be fired.

Remember: any person with authorization to security-related commands is able
to do serious damage.  A truly unscrupulous individual can do just as much
damage with an unauthorized save/restore or use of SST.

This is not meant to be a recommendation against separation of duties.
Instead, i5/OS combined with common sense security practices is safer than
any other common business platform.  However, you have to realize that many
of the industry standard security measures don't apply the same way to - and
are often unneeded with - i5/OS, and the fact that IBM doesn't market this
to the world is a travesty.

Joe