As I understand it CPACF is basically some hardware instructions you can invoke 
from assembler code (I've been using AES128 and SHA1  for our inter-system 
communication software for quite some time). CEXx is a subsystem which can only 
be accessed via various APIs (ICSF). Although  CPACF now supports protected 
keys, this probably in practice requires use of ICSF and a CEX3 facility. While 
one can no doubt load CPACF protected keys oneself it makes little sense to me 
to use protected key without secure key and the feature is more intended to 
improve the efficiency and security of the CEX operations, I guess.  

That said, the CPACF MSA functions  are synchronous, and get executed like any 
other hardware instructions (more or less). Although there is no doubt a little 
bit of setup when the keys are clear text  there's no great overhead. I would 
not expect  the size of blocks to be such a major consideration above a 
sensible minimum size, as seems to be borne out by the white paper. The 
operations are just for symmetric encryption and hash generation (and PRNG).

CEX otoh is accessed via a queuing mechanism. It is asynchronous and suspends 
the executing work unit until the crypto-operation is complete (along with 
encrypting and decrypting keys etc). Obviously this is an enormous overhead 
compared to MSA and the size of data would play a much more significant role. 
Of course it also does a lot more like handling SSL protocols, asymmetric 
encryption and protecting crypto-keys. Costs you more cash as well.  

Interested to know if this is a reasonable summary (if this has all been 
discussed in previous parts of the thread, then sorry).

-----Ursprüngliche Nachricht-----
Von: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] Im Auftrag 
von Greg Boyd
Gesendet: Sonntag, 8. Juli 2012 22:54
An: IBM-MAIN@LISTSERV.UA.EDU
Betreff: Re: Secure Encryption Keys vs Protected Keys

Replying again to finish the last sentence!


I'll point you to the IBM Crypto performance whitepapers, available at 
www.ibm.com/systems/z/advantages/security/z10cryptography.html.  (Look on the 
right under 'Learn More' for your machine type.)  The numbers are very ivory 
tower, and your mileage will vary, however, you can use these numbers to 
compare clear key, secure key and protected key.

Rob:  Protected key does rely on one trip to the Crpyto Express card.  Since a 
protected key begins life as a secure key, the operational key must first be 
decrypted from under the master key (inside the CEX3) but then it is wrapped 
using the wrapping key.  That operational key is then unwrapped inside the 
CPACF hardware.

And as Phil and Rob point out, blocking can have a significnat impact on 
performance.  That's easy to see in the white papers too since they provide 
results for various blocksizes.  The crypto hardware (both CPACF and CEX card) 
is designed to handle large blocks of data and you'll get the best throughput 
with large blocks. So your application design can have a major impact on your 
performance.  

Greg Boyd
IBM Advanced Technical Support
Supporting Crypto on System z

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions, send email to 
lists...@listserv.ua.edu with the message: INFO IBM-MAIN

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN

Reply via email to