HTTP Secure

Hypertext Transfer Protocol Secure (HTTPS) is a combination of the Hypertext Transfer Protocol with the SSL/TLS protocol to provide encrypted communication and secure identification of a network web server. HTTPS connections are often used for payment transactions on the World Wide Web and for sensitive transactions in corporate information systems. HTTPS should not be confused with Secure HTTP (S-HTTP) specified in RFC 2660.

Main idea

The main idea of HTTPS is to create a secure channel over an insecure network. This ensures reasonable protection from eavesdroppers and man-in-the-middle attacks, provided that adequate cipher suites are used and that the server certificate is verified and trusted.

The trust inherent in HTTPS is based on major certificate authorities which come pre-installed in browser software (this is equivalent to saying "I trust certificate authority (e.g. VeriSign/Microsoft/etc.) to tell me whom I should trust"). Therefore an HTTPS connection to a website can be trusted if and only if all of the following are true:

1. The user trusts that their browser software correctly implements HTTPS with correctly pre-installed certificate authorities.
2. The user trusts the certificate authority to vouch only for legitimate websites without misleading names.
3. The website provides a valid certificate (an invalid certificate shows a warning in most browsers), which means it was signed by a trusted authority.
4. The certificate correctly identifies the website (e.g. visiting https://example and receiving a certificate for "Example Inc." and not anything else [see above]).
5. Either the intervening hops on the Internet are trustworthy, or the user trusts the protocol's encryption layer (TLS or SSL) is unbreakable by an eavesdropper.

Browser integration

When connecting to a site with an invalid certificate, older browsers would present the user with a dialog box asking if they wanted to continue. Newer browsers display a warning across the entire window. Newer browsers also prominently display the site's security information in the address bar.

Extended validation certificates turn the address bar green in newer browsers. Most browsers also display a warning to the user when visiting a site that contains a mixture of encrypted and unencrypted content.

Many web browsers, including Firefox (shown here), use the address bar to tell the user that their connection is secure, often by coloring the background.

Most web browsers alert the user when visiting sites that have invalid security certificates. This example is from Firefox.

The Electronic Frontier Foundation, opining that "[i]n an ideal world, every web request could be defaulted to HTTPS", has provided an add-on for the Firefox browser that does so for several frequently used websites.^[1][2]

Technical

Difference from HTTP

As opposed to HTTP URLs which begin with "http://" and use port 80 by default, HTTPS URLs begin with "https://" and use port 443 by default.

HTTP is unsecured and is subject to man-in-the-middle and eavesdropping attacks which can let attackers gain access to website accounts and sensitive information. HTTPS is designed to withstand such attacks and is considered secure against such attacks (with the exception of older deprecated versions of SSL).

Network layers

HTTP operates at the highest layer of the OSI Model, the Application layer; but the security protocol operates at a lower sublayer, encrypting an HTTP message prior to transmission and decrypting a message upon arrival. Strictly speaking, HTTPS is not a separate protocol, but refers to use of ordinary HTTP over an encrypted Secure Sockets Layer (SSL) or Transport Layer Security (TLS) connection.

Server setup

To prepare a web server to accept HTTPS connections, the administrator must create a public key certificate for the web server. This certificate must be signed by a trusted certificate authority for the web browser to accept it. The authority certifies that the certificate holder is indeed the entity it claims to be. Web browsers are generally distributed with the signing certificates of major certificate authorities so that they can verify certificates signed by them.

Acquiring certificates

Authoritatively signed certificates may be free^[3][4] or cost between US$13^[5] and $1,500^[6] per year.

Organizations may also run their own certificate authority, particularly if they are responsible for setting up browsers to access their own sites (for example, sites on a company intranet, or major universities). They can easily add copies of their own signing certificate to the trusted certificates distributed with the browser.

There also exists a peer-to-peer certificate authority, CACert.

Use as access control

The system can also be used for client authentication in order to limit access to a web server to authorized users. To do this, the site administrator typically creates a certificate for each user, a certificate that is loaded into his/her browser. Normally, that contains the name and e-mail address of the authorized user and is automatically checked by the server on each reconnect to verify the user's identity, potentially without even entering a password.

In case of compromised private key

A certificate may be revoked before it expires, for example because the secrecy of the private key has been compromised. Newer versions of popular browsers such as Google Chrome, Firefox,^[7] Opera,^[8] and Internet Explorer on Windows Vista^[9] implement the Online Certificate Status Protocol (OCSP) to verify that this is not the case. The browser sends the certificate's serial number to the certificate authority or its delegate via OCSP and the authority responds, telling the browser whether or not the certificate is still valid.^[10]

Limitations

SSL comes in two options, simple and mutual.

The mutual flavor is more secure but requires the user to install a personal certificate in their browser in order to authenticate themselves.

Whatever strategy is used (simple or mutual), the level of protection strongly depends on the correctness of the implementation of the web browser and the server software and the actual cryptographic algorithms supported. See list in HTTP_Secure#Main idea.

SSL doesn't prevent the entire site from being indexed using a web crawler, and in some cases the URI of the encrypted resource can be inferred by knowing only the intercepted request/response size.^[11] This allows an attacker to have access to the plaintext (the publicly-available static content), and the encrypted text (the encrypted version of the static content), permitting a cryptographic attack.

Because SSL operates below HTTP and has no knowledge of higher-level protocols, SSL servers can only strictly present one certificate for a particular IP/port combination.^[12] This means that, in most cases, it is not feasible to use name-based virtual hosting with HTTPS. A solution called Server Name Indication (SNI) exists which sends the hostname to the server before encrypting the connection, although many older browsers don't support this extension. Support for SNI is available since Firefox 2, Opera 8, and Internet Explorer 7 on Windows Vista.^[13][14][15]

If parental controls are enabled on Mac OS X, HTTPS sites must be explicitly allowed using the Always Allow list.^[16]

From an architectural point of view:

1. An SSL connection is managed by the first front machine which initiates the SSL connection. If, for any reasons (routing, traffic optimization, etc.), this front machine is not the application server and it has to decipher data, solutions have to be found to propagate user authentication informations or certificate to the application server which needs to know who is going to be connected.
2. For SSL with mutual authentication, the SSL session is managed by the first server which initiates the connection. In situations where encryption has to be propagated along chained servers, session timeOut management becomes extremely tricky to implement.
3. With mutual SSL, security is maximal, but on the client-side, there is no way to properly end the SSL connection and disconnect the user except by waiting for the SSL server session to expire or closing all related client applications.
4. For performance reasons, static contents are usually delivered through a non-crypted front server or separate server instance with no SSL. As a consequence, these contents are usually not protected.

History

Netscape Communications created HTTPS in 1994 for its Netscape Navigator web browser.^[17] Originally, HTTPS was used with SSL encryption. As SSL evolved into Transport Layer Security (TLS), the current version of HTTPS was formally specified by RFC 2818 in May 2000.^[18]

See also

• AAA protocol
• Computer security
• Secure Hypertext Transfer Protocol
• Stunnel

References

External links

• RFC 2818: HTTP Over TLS
• SSL 3.0 Specification (IETF)
• Apache-SSL homepage (No longer actively developed)
• Apache 2.2 mod_ssl documentation
• HTTPS Protocol in Internet Explorer Development - MSDN
• Manually Configuring Windows Communication Foundation (WCF) when using HTTP and HTTPS - MSDN
• HTTPS Security Improvements in Internet Explorer 7 & its Compatibility Impact - MSDN
• View this Wikipedia article (and others) with HTTPS/SSL encryption
• curl-loader - HTTPS loading/testing open-source SW