We present a practical methodology for securing the password-based authentication scheme. We propose a solution based on the well-known (k,n) threshold scheme of Shamir for sharing a secret, where in our case the secret is the password itself and (k,n) threshold scheme means that n password-derived secrets (shares) are created and k less than n shares are necessary and sufficient for reconstructing the password, while k-1 are not sufficient. The scheme is information-theoretic secure. Since each of the n shares is stored on a different host (shareholder), an attacker will need to compromise k different shareholders for obtaining an amount of data sufficient for reconstructing the secret. Furthermore, in order to be resistant to the compromising of the server (dealer) coordinating the shareholders we define a variant of the classic Shamir, where the Shamir's abscissae are unknown to dealer and shareholders, making the reconstruction impossible even in the case of dealer and shareholders compromised. In addition we use the Pedersen technique for allowing the verification of shares. For the described scenario we have designed two protocols: the registration (user's sign-up, to be carried out once), and authentication (user's login). We analyse several scenarios where dealer and/or shareholders are partially/totally compromised and confirm that none of them is enabling the attacker to break the authentication. Furthermore we focus on cases where one or more byzantine servers are presented, analysing the impact on the framework and show the adopted mechanisms to be safe against these kinds of attacks. We have developed a prototype demonstrating that our framework works correctly, effectively and efficiently. It provides a first feasibility study that will provide a base for structured and engineered cloud-based implementations aiming at providing an authentication-as-a-service.