[LINK] Why passwords have never been weaker—and crackers have never been stronger

[Martin Barry]

$quoted_author = "Rick Welykochy" ;
> 
> Password cracking is practically infinitely more difficult when a
> password hash is "salted", as mentioned in pages 3 and 4 of the article.
> 
> Rainbow tables are useless with a password salted with 32 random bits.
> 
> Salting has been used by Unix for, oh, at least 25 years. I think Mickeysoft
> finally figured the same thing out after Windows XP and NT.

But the point of the article is that Moore's law has caught up with salting
+ fast hashing algos which makes it sane to just brute force attack rather
than pre-compute rainbow tables. You actually need salting + a *slow*
hashing algo.  

 
> The other improvement is to use the best hashing algorithms currenty
> available, which would be SHA256/512 or better.

My understanding is that current best practice is to use bcrypt which has
salting built in and also has a "work" variable which can be adjusted to
match advances in technology. It slows down the hashing process which is a
small cost for the system but a huge cost for the attacker.


> As the article observes, neither of salting or secure hashing are being
> used by some pretty big players on the web. Many simply store the insecure
> MD5 hash, unsalted. Voila. Rainbow tables make mincemeat of such a scheme
> in seconds.

It's scary how many issues keep cropping up when it seems so easy to get
this right.


> Let's lay the blame on those who can make the difference: those implementing
> the security of the password store. The user can hardly be blamed when
> even large corporations cannot get it right.

Required reading for all: http://codahale.com/how-to-safely-store-a-password/

cheers
Marty