Application layer is the highest level of internet protocal layers.

1. Principles of Network Applications

Network applications are programs that are run on hosts and communicate over network.

1-1. Classic Structure 1: Client Server Architecture (C/S)

• Server offers service to the client,
• Client requests service from the server.

1-2. Classic Structure 2: P2P Architecture

• Example: Bit Torrent
• Peers acts as both the server and the client.
  • A peer downloads one application data from another peer, then simultaneously upload the data to another peer.
  • The more peers in demand for the files, the faster the process is.
  • Self-scalability: new peers bring new service capacity, as well as new service demands.
• However, P2P architecture is a bit harder to manage than the C/S structure.
  • Users may join and leave at anytime,
  • so the P2P software needs to keep track of it.

1- 3. Application Requirements:

• For phone calls, data integrity is not so important,
  - packet loss are still tolerable because we can ask to repeat what they said.
• Throughput can be important to movie streaming apps.
  • Maybe not for movie downloading apps, because it can take as much as it wants and we can work on other things.
• Timing can be important for games (low latency)
• Security can be important for online shopping.

1-4. Application Layer Defines...:

• Open protocols means that it is openly accessible and implementable.

1-5. Transport Layer Protocols: TCP/UDP

• TCP when data should not be lost,
• UDP when data can be lost - like multimedia applications.

2. Web and HTTP

2-1. The Web?:

• Web is one of many applications in the internet.

2-2. HTTP:

• Language used by web server to communicate with clients and web server.
  

2-3. HTTP Over TCP

• On port 80.
• TCP Connection is requested first, then HTTP messages are exchanged. After successful transmission, connection is closed.
• HTTP determines what type of message to be sent, TCP determines how it is transmitted.
  
• t1-t0: transmission delay
• t2-t1: propagation delay
• The TCP connection requests are sent in packets.
• After successful HTTP transaction, the TCP connection is closed.

2-3-1. Timeline for Non-persistent HTTP Connection (http 1.0)

• The browser would initiate a TCP connection request to the server.
• Server accept the TCP connection request.
• Client sends HTTP request packet,
• Server sends back the requested HTML with HTTP response message attached to it.
• Client receives html file, parses it, then notices JPEG object.
• The connection is already closed, so the client does step 1-5 again for requesting JPEG object.
• One object per connection.
• HTTP 1.1 supports multiple objects transmission by not closing the TCP connection after every object transmitted like HTTP 1.0.
  
• RTT (Round Trip Time)
• 2 * RTT + file transmission time

2-4. Persistent HTTP

• HTTP 1.0 (b) is when HTTP requests for objects are sent back-to-back to decrease the time it takes to receive (pipelining).
• HTTP 1.1 (b) also pipelines so that after parsing the html files and noticing the objects, the http requests for objects are sent back-to-back.

• The first two lines of header lines are compulsory.
• keep-alive for HTTP 1.1.
• The last line is blank (\r\n)to indicate the end of header lines.
  
• format of htttp requests.
• cr: carriage return (returns to the start of the line)
• lf: line feed (pulls up to the next line)

• head asks server to get information about a document without retrieving the document itself.
  
  
• curl -I downloads only the header file.
  • The status code is 301 because the requested objects moved from http to https.

HTTP Headers

Cookie

Conditional GET (If-modified-since)

• if there is no modified object, just use locally cached data for the requested objects.