Understanding HMAC authentication and applying it to HTTP

I have been reading up on hash-based message authentication code (HMAC), and how amazon use it to secure their apis.

Why HMAC?

The idea behind HMAC is to generate a cryptographic hash of some data, using a secret key that the client and server both know. The hash generated can then be used by the other party to check the authenticity of the user by performing the same check and comparing the hash, but without both parties revealing the key that was used to sign.

It’s not so much about hiding the contents of a message, it’s more about integrity and proving the user that sent the request must have the same key.

For clarity, it’s worth pointing out that a hash of the same message, hashed with different keys will generate a completely different value, and intercepting a messaged signed by the same user key, but for different messages will also generate a completly different hash, so its theoretically impossible to derive the original key by capturing messages or modifying the request by using say a man in the middle attack.

Lets try it out using a HTTP Request

In order to generate a signature hash from a request, a canonical standardised version of the request must be created so that the same can be performed on the server.

Amazon do this by turning the request into a string in the following format before hashing the string.

According to their documentation, they use this format:

• RequestType
• Domain
• URI
• Sorted string of parameters in property=value format

There isn’t a standard way to do this, so technically as long as the client and server generate a canonicalised version of a request the same way, then we should be able to use it.

Here’s a canonicalised string I want to use for my example (similar to how amazon do it)

GET
craig.goddenpayne.co.uk
/signing-web-requests-using-hmac
SomeHeader=Value&UserId=12345

The Client knows that he is user 12345 (PublicKey) and has a (PrivateKey) of Craig's-Secret-Key

Similarly, the Server knows that user 12345 (PublicKey) will sign requests using the (PrivateKey) Craig's-Secret-Key

I can then hash my message using SHA256 with my PrivateKey, and it would generate a signature such as

13eaaefdfcfb6a66f6d1df27a6458e8925c9f3a4c9480029a60443fad0d4fba9

The signature can then be appended to my request, and now the request I am sending contains the UserId (PublicKey) and the Signature (Canonical message hashed with PrivateKey)

Lets hit the server

GET https://craig.goddenpayne.co.uk/signing-web-requests-using-hmac?SomeHeader=Value&UserId=12345&Signature=13eaaefdfcfb6a66f6d1df27a6458e8925c9f3a4c9480029a60443fad0d4fba9

Its now the job of the server now is do pretty much the same thing to make sure the hash matches.

The server will intepret the request, remove the Signature and creates the canonicalised version of the request

GET
craig.goddenpayne.co.uk
/signing-web-requests-using-hmac
SomeHeader=Value&UserId=12345

The user is looked up using the PublicKey (UserId) and the PrivateKey is used to generate the hash the same way.

If the hash signature matches the signature parameter passed to the server from the client, it can then verify that the request was signed by the same PrivateKey and therefore the request can be considered authentic.