RE: iSeries passwords

["Hall, Philip" <phall@xxxxxxxx>]

> -----Original Message-----
> From: Scott Klement [mailto:klemscot@xxxxxxxxxxxx]
>
> That has absolutely no impact on the CHGPWD command, or the standard
> sign-on screen.  It only affects the "bypass sign-on screen" password
> which is sent before the 5250 data commences.   Even then, it only
> encrypts the password nothing else.

Yes, and I corrected my statement quickly after in a second email...I was 
thinking about the initial negotiation of the session. As *you* know, but I'll 
set it out for those that don't;

 * For vanilla 5250, the password is sent in the clear. 
 * For 5250E the password is encrypted client side before sending.
 * For OpsNav the password is encrypted client side before sending.
 * Using SSL for vanilla 5250 removes it being in the clear.

> 
> > But if you're using the short 10 char passwords, I wouldn't 
> > worry too
> > much about whether it's in the clear or not. The difference to the
> > 'man-in-the-middle' is only a matter of a few minutes.
> 
> Why do you say that?

Again, with regard to the session start-up, if you're using either vanilla 5250 
*or* 5250E (without SSL) the end result for the 'man-in-the-middle' is that;

 * For vanilla 5250, the password is right there for the picking.
 * For 5250E the password is obtainable within a few minutes extra work.
 * For OpsNav the password is obtainable within a few minutes extra work.

Conclusion : For remote access, when security is key, the only safe option for 
now is SSL.

But this isn't a iSeries 'only' issue, on Unix/Linux the default behaviour of 
some of the TCP/IP protocol clients is also to send passwords in the clear.

--phil