Peter Sylvester wrote: >IMO the risk is not to have a non reversible key, but to hide >this "a bit". with a small number of output bits there are >not many options.
Well, if the goal is to protect the data, then it needs to be cryptographically secure. "Security through obscurity" isn't security. <plug> Voltage SecureData is an encryption platform that provides Format-Preserving Encryption, using a mode of AES (almost but not quite through the NIST approvals process-awaiting final comments only) that does what it sounds like: preserves the format of the input. So if you encrypt a 9-digit number, you get back a 9-digit number. If you encrypt "Peter Sylvester, 123 Main St", you get back Xxxxx Xxxxxxxxx, nnn Xxxx Xx", where the Xs and Ns are characters and digits, respectively. And there are lots more options. Using this technology is a great way to mask data: it requires no maintenance-intensive back-end databases (as do traditional masking products) and can be added to an existing procedure as a single extra step. The fact that the data is "obviously" masked (that is, "Peter" is unlikely to encrypt to something that actually LOOKS like a regular name) is actually a plus, since it makes the auditors happy to be able to verify that the data is protected by sampling, rather than having to test or read code. </plug> I'm not a crypto guy, but I know enough to be very uncomfortable with some of the suggestions that have been made on this topic from a security standpoint. This hiding-data-for-test isn't just A Good Idea: it's mandated by various regulations. And given the number of breaches we read about daily, coupled with the (typically) lower security on test systems (or unknown security on partner systems!), it isn't something to take casually: it's playing you-bet-your-company (or at least your job) in many cases. Cheers, -- ...phsiii Phil Smith III p...@voltage.com<mailto:p...@voltage.com> Voltage Security, Inc. www.voltage.com<http://www.voltage.com> (703) 476-4511 (home office) (703) 568-6662 (cell) ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN