Good god, this topic is discussed to death and yet there are spreading very inaccurate, insecure crap in this thread. I think everyone needs to stop giving advice and speculating about what is "good" and defer to a security expert or a set of codified best practices. I just don't understand how some people seem to understand salting, hashes or encryption but not enough to understand why unique-per-user salts are important, why asymmetric and symmetric encryption differ or why encryption has nothing to do with this style of password storage. If you don't understand these things, stop giving security advice in these threads.
>but not enough to understand why unique-per-user salts are important
I don't think that's been covered in this thread. If the main point of salting is to make rainbow tables ineffective, a single DB-wide hash still does that. Presumably if a hacker gets a copy of your entire database, they still don't have a copy of your hashing function with that salt. So then they're reduced to brute force.
Something else that I just thought of is, if you're salting per user, where does that salt get stored? A secondary database?
..And then come to find out it's been discussed down thread. Oops.
Any competent security architect will assume that if the bad guy gets a copy of the entire password database he also gets a copy of the entire codebase, all design documents, installation instructions, and operations manuals, and designs the system to be secure against that.
To whomever downvoted me, I've seen people suggest on this thread or ones in the last 24 hours that:
- symmetric encryption is appropriate for sites' storing passwords
- md5 is okay
- site-wide salts are just as secure as user specific salts
and more. Some of these were suggested by people that have far, far more karma than me. Some were suggested by people that had had this explained to them in previous threads and then took it upon themselves to mis-convey this information back. I don't have a pony in this race as I use different strong passwords per-site and I don't have user data to worry about. I just cringe everytime I see potential hackers reading this misinformation or hackers potentially with data who don't understand their risk or blindspots. I only mean the best, even if I come off exasperated.
I mean sure, if you can get access to one password, you will be able to predict passwords for other websites, however, these SALT + WEBSITE-passwords tend to end up very, very long, which makes it very hard to break. Afterall, every single password you listed there is at least 13 characters long.
