Archive for June, 2012

Security – why programmers should study computing history

You can now add LinkedIn, eHarmony and last.fm to the long list of major sites that have had poor password security in their user database designs.  The saddest part is that in the case of LinkedIn, at least, this was apparently completely avoidable. (I haven’t found enough details to comment on the others, yet.)

Protecting stored user passwords is not rocket science.  This problem was pretty much solved in the 80s and 90s: Use a salted one-way hash function of sufficient strength to resist a dictionary attack.

(LinkedIn’s mistake was to use hashes, but to not salt them. )

That’s it.  Really.  UNIX has been using a salted hash since about 1985, initially with a hash based on DES. Since that time, as computing speeds have increased, new (salted) hash functions based on MD5, Blowfish, and SHA-2 have all been introduced.

In other words, stored password security has been a solved problem for at least 25 years. The concept is the same, only the algorithms have needed to be updated as Moore’s Law has dictated.

This is just one reason that programmers (and sysadmins) should study history, if only the history of computer security. Oh, if you’re not a cryptologist, for security-critical functions, please use well-vetted library functions.

A few references:

Advertisements

, , , , , , ,

Leave a comment

%d bloggers like this: