Preferred method of hashing salted password

18 Aug 2006 3:39 PM

Jono

Hi,
I just discovered the Rfc2898DeriveBytes class, thanks to the MSDN
nugget on storing passwords [1]. The thing is, it left me hanging at
the end. Should I be using this new class or should I be using SHA256
to hash a concatenation of a user's password with a salt generated by
means of an RNGCryptoServiceProvider. Also, if I choose the second
method, is there tangible benefit to creating a random length salt, as
observed here [2]? If anyone has the answers to these questions and
wouldn't mind sharing them, I'd really appreciate it.
Many thanks,
Jono

[1] http://www.microsoft.com/uk/msdn/events/nuggets.aspx
[2] http://www.obviex.com/samples/hash.aspx

18 Aug 2006 4:31 PM

serge calderara

use the keyed hash classes

Show quoteHide quote

"Jono" wrote:

> Hi,
> I just discovered the Rfc2898DeriveBytes class, thanks to the MSDN
> nugget on storing passwords [1]. The thing is, it left me hanging at
> the end. Should I be using this new class or should I be using SHA256
> to hash a concatenation of a user's password with a salt generated by
> means of an RNGCryptoServiceProvider. Also, if I choose the second
> method, is there tangible benefit to creating a random length salt, as
> observed here [2]? If anyone has the answers to these questions and
> wouldn't mind sharing them, I'd really appreciate it.
> Many thanks,
> Jono
>
> [1] http://www.microsoft.com/uk/msdn/events/nuggets.aspx
> [2] http://www.obviex.com/samples/hash.aspx
>
>

19 Aug 2006 10:52 AM

Dominick Baier

Hi,

well - i am a little biased :))

But you should use the Rfc2898 class which gives you the additional benefit
of iterations. 16 bytes lengths are enough.

dominick
www.leastprivilege.com

Show quoteHide quote

> Hi,
> I just discovered the Rfc2898DeriveBytes class, thanks to the MSDN
> nugget on storing passwords [1]. The thing is, it left me hanging at
> the end. Should I be using this new class or should I be using SHA256
> to hash a concatenation of a user's password with a salt generated by
> means of an RNGCryptoServiceProvider. Also, if I choose the second
> method, is there tangible benefit to creating a random length salt, as
> observed here [2]? If anyone has the answers to these questions and
> wouldn't mind sharing them, I'd really appreciate it.
> Many thanks,
> Jono
> [1] http://www.microsoft.com/uk/msdn/events/nuggets.aspx [2]
> http://www.obviex.com/samples/hash.aspx
>

21 Aug 2006 1:31 PM

Jono

I appreciate your response Dominick.
I'll give the Rfc2898 class a try next time I'm given the task of
storing user passwords. I'll admit, I had my doubts about the length of
the hash, but if you reckon 16 bytes is enough (sounds like Gates'
infamous quote) then I'll go along with it.
Cheers,
Jono

Dominick Baier wrote:
Show quoteHide quote

> Hi,
>
> well - i am a little biased :))
>
> But you should use the Rfc2898 class which gives you the additional benefit
> of iterations. 16 bytes lengths are enough.
>
>
> dominick
> www.leastprivilege.com
>
> > Hi,
> > I just discovered the Rfc2898DeriveBytes class, thanks to the MSDN
> > nugget on storing passwords [1]. The thing is, it left me hanging at
> > the end. Should I be using this new class or should I be using SHA256
> > to hash a concatenation of a user's password with a salt generated by
> > means of an RNGCryptoServiceProvider. Also, if I choose the second
> > method, is there tangible benefit to creating a random length salt, as
> > observed here [2]? If anyone has the answers to these questions and
> > wouldn't mind sharing them, I'd really appreciate it.
> > Many thanks,
> > Jono
> > [1] http://www.microsoft.com/uk/msdn/events/nuggets.aspx [2]
> > http://www.obviex.com/samples/hash.aspx
> >