4003-420-01/4005-740-01 Data Communications and Networks I Module 6. Internet Applications -- Lecture Notes Prof. Alan Kaminsky -- Fall Quarter 2012 Rochester Institute of Technology -- Department of Computer Science Internet Requests For Comment (RFCs): http://www.ietf.org/rfc.html Assigned port numbers: http://www.iana.org/assignments/service-names-port-numbers/service-names-port-numbers.xml Remote Connection: Telnet RFC 854   Telnet client Opens a TCP socket connection to some host and port (default port 23) Each character read from the keyboard is written to the socket Each character read from the socket is written to the display   Telnet server Listens for TCP socket connections on the well-known Telnet port number, 23 Each socket is handed off to a login process   Security Telnet is insecure -- anyone can sniff the data going back and forth over the socket connection Secure Shell (SSH) is an alternative that is secure   The Telnet client can connect to any port, not just port 23 This is useful for learning about Internet protocols, those that use textual encoding of messages The Web: Hypertext Transfer Protocol (HTTP) Architecture Web servers, web clients (browsers) HTTP requests, HTTP responses TCP transport, textual wire protocol Pull model -- client pulls data from server The difference between HTTP and HTML   Standards RFC 1945: HTTP/1.0 (also HTTP/0.9) RFC 2616: HTTP/1.1   HTTP request message format Method, request-URI, HTTP-version HTTP GET request From browser URL to TCP connection and HTTP GET request message Demo: Telnet to localhost web server (↵ stands for the return key) $ telnet localhost 80↵ GET /~ark/ HTTP/1.0↵ ↵   HTTP response message format HTTP-version, status code, reason phrase Header fields Entity body Demo: Try to get a nonexistent web page $ telnet localhost 80↵ GET /~ark/garbage.html HTTP/1.0↵ ↵   Web browser operation Get main object (URL) Decide how to display page Get embedded objects (URLs)   Web server operation HTTP is stateless on the server side   Types of connections Nonpersistent connections (HTTP/1.0, HTTP/1.1) Each HTTP request/response is carried over a separate connection Each HTTP request/response is performed one at a time Persistent connections without pipelining (HTTP/1.1) Each HTTP request/response to a certain server is carried over the same connection Each HTTP request/response to this server is performed one at a time Persistent connections with pipelining (HTTP/1.1) Each HTTP request/response to a certain server is carried over the same connection The HTTP request/response for the main document is performed first Any HTTP requests for additional referenced documents (e.g., images) are performed all at once Multiple simultaneous connections (HTTP/1.0, HTTP/1.1) Each HTTP request/response to a certain server is carried over a separate connection The HTTP request/response for the main document is performed first Any HTTP requests/responses for additional referenced documents (e.g., images) are performed all at once Number of round trip times (RTTs) needed to retrieve a web page with each type of connection Persistent connections demo $ telnet localhost 80↵ GET /~ark/ HTTP/1.1↵ Host: localhost↵ ↵ (main web page comes back) GET /~ark/fire.html HTTP/1.1↵ Host: localhost↵ ↵ (fire.html web page comes back) Determining the end of each HTTP response message with persistent connections Can no longer rely on socket input stream end-of-file to mark the end of the HTTP response message Alternative: Content-Length header (/~ark/fire.html) Alternative: Chunked transfer coding (/~ark/)   Web page caching Conditional GET with If-Modified-Since: header Demo: Telnet to localhost web server, get a .html page $ telnet localhost 80↵ GET /~ark/menubar.html HTTP/1.0↵ ↵ Demo: Try again, include the If-Modified-Since: header $ telnet localhost 80↵ GET /~ark/menubar.html HTTP/1.0↵ If-Modified-Since: Tue, 07 Jan 2003 12:00:00 GMT↵ ↵ Caching in web clients Caching in web cache servers   Session state: Cookies RFC 2109 Demo: Use wget to retrieve www.google.com home page and print HTTP headers wget -S http://www.google.com/ How cookies are used to track state within one web server Server sets a cookie in an HTTP response message header Thereafter, browser sends the cookie in an HTTP request message header whenever the browser contacts that server again User preferences Shopping carts How cookies are used to track state across multiple web servers Banner ads Privacy issues   Web page forms Sending form data with HTTP GET requests Example: www.google.com, main page Sending form data with HTTP POST requests Susceptible to eavesdropping Secure HTTP (https:) -- encrypts data sent and received Internet Email Architecture Mail servers, outgoing mail queues, incoming user mailboxes User agents Protocol for delivering email to user mailboxes: SMTP Protocol for retrieving email from user mailboxes: POP3, IMAP   Simple Mail Transfer Protocol (SMTP) RFC 2821 TCP based, textual protocol, port 25 Push model -- client pushes data to server (vs. HTTP's pull model) Demo client program $ java SmtpDemo smtp-server.rit.edu 25 kaminsky.cs.rit.edu \ 'ark@cs.rit.edu' 'arkics@rit.edu' 'Test' 'This is a test' 220 mxgate03.rit.edu ESMTP RIT Mail Gateway HELO kaminsky.cs.rit.edu 250 mxgate03.rit.edu MAIL FROM: <ark@cs.rit.edu> 250 Ok RCPT TO: <arkics@rit.edu> 250 Ok DATA 354 End data with <CR><LF>.<CR><LF> Date: Thu, 8 Oct 2009 11:45:59 -0400 From: ark@cs.rit.edu To: arkics@rit.edu Subject: Test This is a test . 250 Ok: queued as 154B7514005 QUIT 221 Bye RFC 822: Email header lines (Date:, From:, To:, etc.) Email ethics SMTP does not authenticate the sender You can put whatever you want after HELO and FROM Enormous potential for abuse (spam) Defense Against the Dark Arts   Multipurpose Internet Mail Extensions (MIME) RFC 2045, RFC 2046 Defines how to transfer non-ASCII data Due to a historical quirk, SMTP only transfers 7-bit ASCII data Non-ASCII data must be encoded in the form of ASCII Quoted printable encoding Base 64 encoding Defines how to interpret non-ASCII data "MIME media types" or "MIME types" Also used in HTTP text/plain: Plain text text/html: Web page image/jpeg: Digital photo image/png: Portable Network Graphics image application/pdf: Adobe Acrobat file application/postscript: PostScript file application/x-ark-encrypted-grade-file: Nonstandard type Defines how to include email attachments MIME type multipart/mixed Example from my inbox   Post Office Protocol (POP3) RFC 1939 TCP based, textual protocol, port 110 Pull model -- client pulls data from server Very simple: list messages, retrieve messages, delete messages The user agent stores messages, manages mail folders, etc. Most useful for a user who always uses the same machine to access email   Internet Mail Access Protocol (IMAP) RFC 2060 Many more features than POP3 Maintains a hierarchy of mail folders on the server side Useful for a user who uses different machines to access email Internet Host Name Directory: Domain Name System (DNS) RFC 1034, RFC 1035   UDP based, binary protocol, port 53   Architecture Local name server Each host is configured with the IP addresses (not host names!) of one or more local name servers Host queries local name server to find the IP address for a host name Root name servers 13 scattered around the Internet If the local name server can't answer the query, it delegates to a root name server List of root name servers: http://www.root-servers.org Authoritative name servers One or more for each domain (like rit.edu) Root name server delegates to an authoritative name server Intermediate name servers Some domains may have additional name servers Root name server delegates to an intermediate name server, which eventually delegates to an authoritative name server Query caching   DNS client program host [options] name [server] host -v ... -- Verbose output host -r ... -- Don't do recursive queries   DNS functions -- demos Host name to IP address mapping -- for mapping URLs host www.cs.rit.edu host -v www.cs.rit.edu IP address to host name mapping host 216.239.51.101 Domain name to authoritative name server mapping host -t NS cs.rit.edu host -t NS rit.edu Domain name to mail server name mapping -- for mapping email addresses host -t MX cs.rit.edu host -v -t MX cs.rit.edu Gives mail server host name plus a numerical preference value (lower values are more preferred) Load balancing host -t MX rochester.rr.com host -t MX rochester.rr.com Data Communications and Networks I • 4003-420-01/4005-740-01 • Fall Quarter 2012 Course Page Alan Kaminsky • Department of Computer Science • Rochester Institute of Technology • 4486 + 2220 = 6706 Home Page Copyright © 2012 Alan Kaminsky. All rights reserved. Last updated 08-Oct-2012. Please send comments to ark­@­cs.rit.edu.