SMU CSE 4344

application layer

SMU CSE 4344

application vs. application-layer protocols

• application-layer protocol is just one piece– how the end hosts communicate

• example: World Wide Web– HyperText Transfer Protocol (HTTP) is protocol– but the Web includes other components, such as

document formats (HTML), Web browsers, servers, …

• example: electronic mail– Simple Mail Transfer Protocol (SMTP) is protocol– but e-mail includes other components, such as mail

servers, user mailboxes, mail readers

SMU CSE 4344

DNS – Domain Name System

• DNS maps ASCII host names (e.g., marge.engr.smu.edu) & email addresses (e.g., donm@engr.smu.edu) to IP addresses

• resolver procedure queries DNS name server with ASCII name parameter, gets back resource record

• numeric IP addresses used for Internet routing• domains are hierarchical

SMU CSE 4344

zones

• DNS name space comprises zones

• each zone has its own name server(s)

• authoritative record vs. cached record

SMU CSE 4344

DNS zones

Part of the DNS name space showing the division into zones.

SMU CSE 4344

DNS lookup illustration

• a resolver looks up host linda.cs.yale.edu• recursive query

SMU CSE 4344

email: architecture and services• user agent• mail transfer agent (daemon)• basic application functions

– composition• addresses, other header fields, text body

– transfer (MX to MX)– reporting (bounced, delivered, still trying)– displaying (folders, messages)

• status: unread, read, answered, deleted• text, attachments

– disposition (delete, save, organize, recycle)

SMU CSE 4344

email message format – RFC 822

RFC 822 header fields related to message transport

SMU CSE 4344

RFC 822

more RFC 822 message header fields

SMU CSE 4344

MIME headers

MIME defines encoding rules for non-ASCII, using five additional RFC 822-compliant headers

SMU CSE 4344

message transfer

• SMTP: Simple Mail Transfer Protocol– TCP over well-

known port 25– ASCII protocol

SMU CSE 4344

final delivery

-Post Office Protocol v. 3 (POP3): client downloads messages (port 110)

- Internet Message Access Protocol (IMAP): msgs stay on server (port 143)

SMU CSE 4344

POP3 vs. IMAP

SMU CSE 4344

World Wide Web (WWW)

• architectural overview• HTTP – HyperText Transfer Protocol• performance enhancements

SMU CSE 4344

WWW origins

• hypertext, Vannevar Bush, M.I.T., 1945• Tim Berners-Lee, CERN, 1989

– proposal: networked “web” of hyperlinked documents

• Marc Andreessen, U Illinois– Mosaic graphical browser, 1993– Netscape, 1994

• World Wide Web Consortium (W3C)– CERN, M.I.T.– standardization effort– director, Tim Berners-Lee

SMU CSE 4344

WWW architecture

• web pages and hyperlinks• web servers host web pages• client software: web browser• basic operation

– user selects a hyperlink– browser requests IP address from DNS– browser requests hyperlink-identified web page – web server sends new web page– browser “renders” new web page– only whole pages sent

SMU CSE 4344

architectural overview

SMU CSE 4344

WWW clients

• URL: Uniform Resource Locator• hyperlinks are based on URLs

– <protocol>://<DNSname>/<filename>– http://engr.smu.edu/~donm/4344.html

• WWW services at well-known port 80• HyperText Markup Language: HTML

SMU CSE 4344

URLs – Uniform Resource Locaters

Some common URLs.

SMU CSE 4344

display of non-HTML content

• plug-ins extend the browsers capabilities• helpers are independent processes• dynamic content: PHP, JSP, ASP, perl, javascript• Wireless Application Protocol (WAP); i-mode; ...

SMU CSE 4344

web servers

multi-threaded Web server with front end and processing modules

SMU CSE 4344

server farm

SMU CSE 4344

caching

Hierarchical caching with three proxies.

SMU CSE 4344

caching vs. replication

• motivations for moving content close to users– reduce latency for user– reduce load on network and server

• caching– replicating content “on demand” after request– storing response message locally for future use

• replication– planned replication of content in multiple locations– updating of resources is handled outside of HTTP– can replicate scripts that create dynamic responses

SMU CSE 4344

caching vs. replication• caching first seen as very important in HTTP

– many additions to HTTP to support caching• in particular, cache validation

• deployment of caching proxies in the 1990s– service providers & enterprises deployed proxies– … to cache content across a community of users– sometimes, gains were not very dramatic

• then, content distribution networks emerged– companies (like Akamai) that replicate Web sites– host all (or part) of a Web site for a content provider– place replicas all over world on many machines

SMU CSE 4344

TCP interaction: multiple transfers• most Web pages have multiple objects

– e.g., HTML file, multiple embedded images

• serializing the transfers is not efficient– sending images one at a time introduces delay– cannot retrieve another image until first arrives

• parallel connections– browser opens multiple TCP connections … and

retrieves a single image on each connection

• performance trade-offs– multiple downloads share same network links– unfairness to other traffic traversing the links

SMU CSE 4344

TCP interaction: short transfers

• HTTP transfers are short– very small request message

(e.g., a few hundred bytes)– small response message

(e.g., a few kilobytes)

• TCP overhead may be big– three-way handshake to

establish connection– four-way handshake to tear

down the connection

time to transmit file

initiate TCPconnection

RTT

requestfile

RTT

filereceived

time time

SMU CSE 4344

TCP interaction: persistent connections

• handle multiple transfers per connection– added to HTTP after Web became very popular– maintain TCP connection across multiple requests– either client or server can tear down the connection

• performance advantages– avoid overhead of connection set-up and tear-down– allow TCP to learn a more accurate RTT estimate– allow the TCP congestion window to increase

• further enhancement: pipelining– send multiple requests one after the other … before

receiving the first response

SMU CSE 4344

reflecting on application-layer protocols

• protocols are tailored to applications– each protocol is customized to a specific need

• protocols have many key similarities– each new protocol was influenced by the previous ones– new protocols commonly borrow from the older ones

• protocols depend on same underlying substrate– Domain Name System (DNS)– e.g., ordered reliable stream of bytes (i.e., TCP)

• relevance of the protocol standards process– important for interoperability across implementations– yet, not necessary if same party writes all of the software

• …which is increasingly common (e.g., P2P software)