The necessity of hiding the salt for a hash

Hiding a salt is unnecessary.

A different salt should be used for every hash. In practice, this is easy to achieve by getting 8 or more bytes from cryptographic quality random number generator.

From a previous answer of mine:

Salt helps to thwart pre-computed dictionary attacks. Suppose an attacker has a list of likely passwords. He can hash each and compare it to the hash of his victim's password, and see if it matches. If the list is large, this could take a long time. He doesn't want spend that much time on his next target, so he records the result in a "dictionary" where a hash points to its corresponding input. If the list of passwords is very, very long, he can use techniques like a Rainbow Table to save some space. However, suppose his next target salted their password. Even if the attacker knows what the salt is, his precomputed table is worthless—the salt changes the hash resulting from each password. He has to re-hash all of the passwords in his list, affixing the target's salt to the input. Every different salt requires a different dictionary, and if enough salts are used, the attacker won't have room to store dictionaries for them all. Trading space to save time is no longer an option; the attacker must fall back to hashing each password in his list for each target he wants to attack. So, it's not necessary to keep the salt secret. Ensuring that the attacker doesn't have a pre-computed dictionary corresponding to that particular salt is sufficient.

After thinking about this a bit more, I've realized that fooling yourself into thinking the salt can be hidden is dangerous. It's much better to assume the salt cannot be hidden, and design the system to be safe in spite of that. I provide a more detailed explanation in another answer.

However, recent recommendations from NIST encourage the use of an additional, secret "salt" (I've seen others call this additional secret "pepper"). One additional iteration of the key derivation can be performed using this secret as a salt. Rather than increasing strength against a pre-computed lookup attack, this round protects against password guessing, much like the large number of iterations in a good key derivation function. This secret serves no purpose if stored with the hashed password; it must be managed as a secret, and that could be difficult in a large user database.

The answer here is to ask yourself what you're really trying to protect from? If someone has access to your database, then they have access to the encrypted salts, and they probably have access to your code as well. With all that could they decrypt the encrypted salts? If so then the encryption is pretty much useless anyway. The salt really is there to make it so it isn't possible to form a rainbow table to crack your entire password database in one go if it gets broken into. From that point of view, so long as each salt is unique there is no difference, a brute force attack would be required with your salts or the encrypted salts for each password individually.

A hidden salt is no longer salt. It's pepper. It has its use. It's different from salt.

Pepper is a secret key added to the password + salt which makes the hash into an HMAC (Hash Based Message Authentication Code). A hacker with access to the hash output and the salt can theoretically brute force guess an input which will generate the hash (and therefore pass validation in the password textbox). By adding pepper you increase the problem space in a cryptographically random way, rendering the problem intractable without serious hardware.

For more information on pepper, check here.

See also hmac.

My understanding of "salt" is that it makes cracking more difficult, but it doesn't try to hide the extra data. If you are trying to get more security by making the salt "secret", then you really just want more bits in your encryption keys.

The second approach is only slightly more secure. Salts protect users from dictionary attacks and rainbow table attacks. They make it harder for an ambitious attacker to compromise your entire system, but are still vulnerable to attacks that are focused on one user of your system. If you use information that's publicly available, like a telephone number, and the attacker becomes aware of this, then you've saved them a step in their attack. Of course the question is moot if the attacker gets your whole database, salts and all.

EDIT: After re-reading over this answer and some of the comments, it occurs to me that some of the confusion may be due to the fact that I'm only comparing the two very specific cases presented in the question: random salt vs. non-random salt. The question of using a telephone number as a salt is moot if the attacker gets your whole database, not the question of using a salt at all.

... something like a user name or phone number to salt the hash. ... My question is if the second approach is really necessary? I can understand from a purely theoretical perspective that it is more secure than the first approach, but what about from a practicality point of view?

From a practical point of view, a salt is an implementation detail. If you ever change how user info is collected or maintained – and both user names and phone numbers sometimes change, to use your exact examples – then you may have compromised your security. Do you want such an outward-facing change to have much deeper security concerns?

Does stopping the requirement that each account have a phone number need to involve a complete security review to make sure you haven't opened up those accounts to a security compromise?

Here is a simple example showing why it is bad to have the same salt for each hash

Consider the following table

UserId  UserName,   Password
     1  Fred       Hash1 =  Sha(Salt1+Password1)    
     2  Ted        Hash2 =  Sha(Salt2+Password2)    

Case 1 when salt 1 is the same as salt2 If Hash2 is replaced with Hash1 then user 2 could logon with user 1 password

Case 2 when salt 1 not the same salt2 If Hash2 is replaced with Hash1 then user2 can not logon with users 1 password.

There are two techniques, with different goals:

The "salt" is used to make two otherwise equal passwords encrypt differently. This way, an intruder can't efficiently use a dictionary attack against a whole list of encrypted passwords.

The (shared) "secret" is added before hashing a message, so that an intruder can't create his own messages and have them accepted.

I tend to hide the salt. I use 10 bits of salt by prepending a random number from 1 to 1024 to the beginning of the password before hashing it. When comparing the password the user entered with the hash, I loop from 1 to 1024 and try every possible value of salt until I find the match. This takes less than 1/10 of a second. I got the idea to do it this way from the PHP password_hash and password_verify. In my example, the "cost" is 10 for 10 bits of salt. Or from what another user said, hidden "salt" is called "pepper". The salt is not encrypted in the database. It's brute forced out. It would make the rainbow table necessary to reverse the hash 1000 times larger. I use sha256 because it's fast, but still considered secure.

Really, it depends on from what type of attack you're trying to protect your data.

The purpose of a unique salt for each password is to prevent a dictionary attack against the entire password database.

Encrypting the unique salt for each password would make it more difficult to crack an individual password, yes, but you must weigh whether there's really much of a benefit. If the attacker, by brute force, finds that this string:

Marianne2ae85fb5d

hashes to a hash stored in the DB, is it really that hard to figure out what which part is the pass and which part is the salt?