Is knowing the decrypted and encrypted data enough to find the key?

Question

Let's say an attacker knows the encrypted content, the decrypted content and the algorithm used: can he get the key that was used to encrypt the content or does it make it easier to find the key out?

I think that this may vary between symmetric and asymmetric encryption and between different algorithms. So how secure is this scenario with different algorithms if the only thing I need to keep secret is the key?

---

EDIT:

I ask this question because I just read through the SpiderOak FAQ and as far as I understood that, they are neither storing a user's password nor the encryption key which is derivered from the user's password. However, they need to make sure that the user's password is the right one when he's logging in to the web page.

If they don't store the password and don't store the key, the only possibility to check whether the password is right is to decrypt some known encrypted data and check whether the result is the right one.

E.g. when registering they might encrypt the data CHECK and when logging in they might derive a key from the entered password and decrypt the encrypted CHECK and see whether it is CHECK. If not, the key was not right, and thus the password wasn't either.

Is there any other way how they could do it?

Answer 1

What you are referring to is a known-plaintext attack, where you attempt to figure out the key by comparing the encrypted and unencrypted versions of a sample of data. Fortunately, pretty much all major encryption algorithms in use today, including AES, are not known to be susceptible to this kind of attack. Knowing what something decrypts to is completely useless as far as figuring out the key goes (unless, perhaps, your approach is to attempt to brute-force the key.)

While Spideroak technically could use the method you describe (i.e, encrypting a known sample file), they actually use a procedure known as hashing according to their website. Hashing is a very common password storage technique; in fact it is used by nearly all websites that process user logins. It's is a mathematical process that takes in some data (such as a password) as input, and generates a unique, fixed-length string as output. It is easy to calculate the hash of some data, but given the hash, it is very hard to figure out what the original data was. But as long as you provide the same data, the hash will always come out the same.

According to Spideroak, when you create an account, your browser calculates the bcrypt hash of your password and sends the hash to spideroak. Presumably, when you log in, spideroak sends the hash back to your browser, and your browser will take the hash of whatever you typed in to check that it matches.