Ed Jaffe recommended against creating a SAF class. I'll respectfully suggest 
that it's not that hard.

First, if you do, IBM told us, "Start the class name with a dollar sign-we'll 
never use those". Of course you could collide with
another vendor, but that's unlikely.

We've had customers doing so for 13 years or so. Besides some folks who didn't 
understand how to use their own ESM, we've had no
problems. ACF2 and TSS were easy, too.

Now, I admit that our usage is pretty simple: we have named data protection 
entities called Cryptids, and you can use them to
protect (encrypt/tokenize/hash) or access (decrypt/detokenize) data. So if you 
have a Cryptid named BANANA, a user needs READ or
greater authority to PROTECT.BANANA or ACCESS.BANANA, as appropriate to use 
BANANA to protect or access. 

For something like EJES, with possibly dozens of subtleties, it would surely be 
harder. The complexity of SAF related to
certificates comes to mind, though I suspect some of that is due to some 
historical mistakes. Still, once you've defined a scheme,
it's just PERMITs, right?


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

Reply via email to