Overview of Hypertext Transfer Protocol

HTTP/1.1

HTTP/1.1 is the current, stable version of Hypertext Transfer Protocol as defined in RFC 2616 of the Internet Engineering Task Force (June 1999).

I. Request/Response Client/Server Protocol

HTTP operates through the exchange of messages:

• HTTP Requests sent from clients to servers.
• HTTP Responses from servers to clients.

Statelessness

HTTP is fundamentally a stateless protocol.

• There are no mandated server-side state changes through a request/response interchange.
• Although a client may interact with a server through a series of interchanges, each such interchange is considered a complete and independent transaction from the HTTP perspective.
• Usually, there are no mandated client-side state changes through an interchange. However, cookies, redirections and actions of scripts can require client-side actions.

Intermediaries: Proxies, Tunnels, Gateways

HTTP messages may go through a number of intermediaries. Requests and responses may be modified along the way.

Proxy

• Acts as both HTTP client and server.
• Receives requests from clients and passes them on, possibly translated.
• Accepts responses from servers and relays them back to originating client.
• Possibly non-transparent: adding information, value added service.

Tunnel

• An intermediary that acts as a blind relay: no modification of requests or responses.

Gateway

• A server which acts as an intermediary for some other server. Also known as a reverse proxy.

Connections

Under HTTP/1.0, a separate TCP connection was generally established for each request/response interchange.

HTTP/1.1 permits and encourages continuous connections for a series of request/response interchanges. This increases communication efficiency.

Pipelining is also supported: clients may send multiple requests without waiting for responses.

II. Requests

The basic structure of an HTTP request is composed of five elements.

• A method (e.g., GET or POST).
• A requested URI (Uniform Resource Identifier) for query/retrieval.
• The HTTP version (HTTP/1.1 typically).
• Headers that may control the request.
• A possible message body (e.g. for POST).

GET Requests

The GET request is the basic method to retrieve content associated with a URI.

• A conditional GET may be issued to retrieve data only under certain circumstances, using headers such as If-Modified-Since, If-Match.
• A partial GET may be issued to retrieve part of a data entity through the Range header.

HEAD Requests

The HEAD request is a partial GET that retrieves only the response headers associated with request. No message body is to be returned.

POST Requests

The POST request is the basic mechanism for uploading data to be associated with a particular URI. It is often used with HTML forms to send data from the client to the server.

PUT and DELETE Requests

HTTP also provides two lesser used requests related to modification of web content. Most web services do not support these methods.

• The PUT request provides new content to be associated with a URI.
• The DELETE request asks that the content associated with a URI be deleted.

Other Methods: OPTIONS, TRACE, CONNECT

The OPTIONS, TRACE and CONNECT methods support various control mechanisms associated with the communication process.

These methods are not typically invoked by user actions.

III. Responses

The basic structure of an HTTP response is composed of four elements.

• The HTTP version (HTTP/1.1 typically).
• A status code and description.
• Response headers.
• The message body of the response

Response Codes

Response codes are organized into five classes.

• 1xx: Informational codes (processing continues).
• 2xx: Success codes.
• 3xx: Redirection codes.
• 4xx: Client error codes (bad request syntax, etc.)
• 5xx: Server error codes.

Redirection

Redirection is a crucial feature of HTTP that requests the client to automatically reissue a request to a different URI.

Time delays can be inserted with the Retry-After header.

Content-Type

Content-type is a key header that tells the client what type of content the server is sending back.

• text/html - HTML content.
• application/octet-stream - binary file.