Transport Layer Security

Secure Sockets Layer (SSL) and Transport Layer Security (TLS), its successor, are cryptographic protocols which provide secure communications on the Internet. There are slight differences between SSL 3.0 and TLS 1.0, but the protocol remains substantially the same. The term "SSL" as used here applies to both protocols unless clarified by context.

Description

SSL provides endpoint authentication and communications privacy over the Internet using cryptography. In typical use, only the server is authenticated (i.e. its identity is ensured) while the client remains unauthenticated; mutual authentication requires public key infrastructure (or PKI) deployment to clients. The protocols allow client/server applications to communicate in a way designed to prevent eavesdropping, tampering, and message forgery.

SSL involves a number of basic phases: Peer negotiation for algorithm support
Public key encryption-based key exchange and certificate-based authentication
Symmetric cipher-based traffic encryption

During the first phase, the client and server negotiate which cryptographic algorithms will be used. Current implementations support the following choices: for public-key cryptography: RSA, Diffie-Hellman, DSA or Fortezza;
for symmetric ciphers: RC2, RC4, IDEA, DES, Triple DES or AES;
for one-way hash functions: MD5 or SHA.

How it works

The SSL protocol exchanges records; each record can be optionally compressed, encrypted and packed with a message authentication code (MAC). Each record has a content_type field that specifies which upper level protocol is being used. When the connection starts, the record level encapsulates another protocol, the handshake protocol, which has content_type 22. The client sends and receives several handshake structures:

It sends a ClientHello message specifying the list of cipher suites, compression methods and the highest protocol version it supports. It also sends random bytes which will be used later.
Then it receives a ServerHello, in which the server chooses the connection parameters from the choices offered by the client earlier.

When the connection parameters are known, client and server exchange certificates (depending on the selected public key cipher). These

certificates are currently X.509, but there's also a draft specifying the use of OpenPGP based certificates.

The server can request a certificate from the client, so that the connection can be mutually authenticated..

Client and server negotiate a common secret called "master secret", possibly using the result of a Diffie-Hellman exchange, or simply encrypting a secret with a public key that is decrypted with the peer's private key. All other key data is derived from this "master secret" (and the client- and server-generated random values), which is passed through a carefully designed "Pseudo Random Function".

TLS/SSL have a variety of security measures:

Numbering all the records and using the sequence number in the MACs.

Using a message digest enhanced with a key (so only with the key can you check the MAC). This is specified in RFC 2104).

Protection against several known attacks (including man in the middle attacks), like those involving a downgrade of the protocol to previous (less secure) versions, or weaker cipher suites.

The message that ends the handshake ("Finished") sends a hash of all the exchanged data seen by both parties.

The pseudo random function splits the input data in 2 halves and processes them with different hashing algorithms (MD5 and SHA), then XORs them together. This way it protects itself in the event that one of these algorithms is found vulnerable.

Applications

SSL runs on layers beneath application protocols such as HTTP, SMTP and NNTP and above the TCP transport protocol, which forms part of the TCP/IP protocol suite. While it can add security to any protocol that uses reliable connections (such as TCP), it is most commonly used with HTTP to form HTTPS. HTTPS is used to secure World Wide Web pages for applications such as electronic commerce. It uses public key certificates to verify the identity of endpoints.

While an increasing number of client and server products can support SSL natively, many still do not. In these cases, a user may wish to use standalone SSL products like Stunnel to provide encryption. However, the Internet Engineering Task Force recommended in 1997 that application protocols offer a way to upgrade to TLS from a plaintext connection, rather than use a separate port for encrypted communications - this prevents use of wrappers such as Stunnel.

SSL can also be used to tunnel an entire network stack to create a VPN, as is the case with OpenVPN.

History and development

Developed by Netscape, SSL version 3.0 was released in 1996, which later served as a basis to develop TLS version 1.0, an IETF standard protocol first defined in RFC 2246. Visa, MasterCard, American Express and many leading financial institutions have endorsed SSL for commerce over the Internet.

SSL operates in modular fashion: its authors designed it for extendability, with support for forwards and backwards compatibility and negotiation between peers.

Early weak keys Some early implementations of SSL could use a maximum of only 40-bit symmetric keys because of US government restrictions on the export of cryptographic technology. The US government explicitly imposed a 40-bit keyspace small enough to be broken by brute-force search by law enforcement agencies wishing to read the encrypted traffic, while still presenting obstacles to less-well-funded attackers. A similar limitation applied to Lotus Notes in export versions. After several years of public controversy, a series of lawsuits, and eventual US government recognition of changes in the market availability of 'better' cryptographic products produced outside the US, the authorities relaxed some aspects of the export restrictions. The 40-bit key size limitation has mostly gone away. Modern implementations use 128-bit (or longer) keys for symmetric key ciphers.

Incorrect uses Some websites have been criticized for incorrectly using SSL and therefore negating the security benefits [1]. Such incorrect uses include:

Only securing the form submission page, while failing to secure the login page

Displaying a secure page mixed with non-secure media

This practice has been found present in many commercial websites such as those of Bank of America and Washington Mutual [4].

Standards

The first definition of TLS appeared in RFC 2246: "The TLS Protocol Version 1.0". Other RFCs subsequently extended TLS, including:

RFC 2712: "Addition of Kerberos Cipher Suites to Transport Layer Security (TLS)". The 40-bit ciphersuites defined in this memo appear only for the purpose of documenting the fact that those ciphersuite codes have already been assigned.

RFC 2817: "Upgrading to TLS Within HTTP/1.1", explains how to use the Upgrade mechanism in HTTP/1.1 to initiate Transport Layer Security (TLS) over an existing TCP connection. This allows unsecured and secured HTTP traffic to share the same well known port (in this case, http: at 80 rather than https: at 443).

RFC 2818: "HTTP Over TLS", distinguishes secured traffic from insecure traffic by the use of a different 'server port'.

RFC 3268: "AES Ciphersuites for TLS". Adds Advanced Encryption Standard (AES) ciphersuites to the previously existing symmetric ciphers. RFC 3546: "Transport Layer Security (TLS) Extensions", adds a mechanism for negotiating protocol extensions during session initialisation and defines some extensions.

RFC 4279: "Pre-Shared Key Ciphersuites for Transport Layer Security (TLS)", adds three sets of new ciphersuites for the TLS protocol to support authentication based on pre-shared keys.