CSE 190: Internet Commerce

1
CSE 190 Internet Commerce

• Lecture 4 Web Servers

2
Web Servers

• What do they do?
• How do they work?
• Apache Installation, Configuration
• Apache Administration
• Special Topics

3
What does a web server do?

• A web server serves files to clients (such as the
  browser)
• Files may be HTML, GIFs, video, PDF
• Serves multiple clients at the same time
• Transfer protocol HTTP

4
Commercial Implementations

• Apache
• IIS
• NetscapeCommerceServer

Netcraft data for active servers
(http//www.netcraft.com/survey/)
5
HTTP

• Simple text command-response protocol
• Always uses TCP/IP (connection oriented)
• Methods GET, POST, HEAD
• Stateless protocol no client identifier
• Uses 8 bit bytes (octets)
• HTTP Versions 0.9 (assumes HTML), 1.0
  (client/server headers), 1.1 (persistent
  connections)
• HTTP status codes
• 200s (Success)
• 300s (Redirect)
• 400s (Bad request)
• 500s (Server Failure)

6
HTTP Illustrated

• GET / HTTP/1.0
• Standard HTTP/1.0 request for main site page
• GET /ex1.html HTTP/1.0
• Standard request for the ex1.html page
• GET /ex1.gif HTTP/1.0
• Request an image file (image/gif)
• GET /notthere.html HTTP/1.0
• Returns 404 (Not found) response
• GET /
• An HTTP/0.9 request
• GET / HTTP/1.1Hostname gremlin.ucsd.eduGET
  /ex1.html HTTP/1.1Hostname gremlin.ucsd.edu
• A pipelined request

7
HTTP Illustrated (POST, HEAD)

• HEAD / HTTP/1.0
• Just view what headers are returned by GET
• POST /cgi-bin/ex2.pl HTTP/1.0Content-Length
  25userguestpasssecret
• Posts two parameters to a script

8
HTTP Illustrated (Realistic request)

• GET /x.html HTTP/1.1Connection
  Keep-AliveAccept image/gif, image/x-xbitmap,
  image/jpeg, image/pjpeg, application/vnd.ms-excel,
  application/vnd.ms-powerpoint,
  application/msword, /Accept-Encoding gzip,
  deflateAccept-Language en-usHost
  gremlin.ucsd.edu60846User-Agent Mozilla/4.0
  (compatible MSIE 6.0 Windows NT 5.0)Cookie
  bookcookie

9
How does a web server work?

• Serial request model (Example printhttp.pl)
• while (true)
• accept client connection
• read HTTP request
• parse request
• pathname req.uri_details
• send HTTP response header
• foreach byte in file pathname
• send byte to client
• close client connection
• Pitfall Requests from other clients have to wait

10
How does a web server work?

• Multiple clients (Apache model)
• Parent
• Listen on Port (server default 80)
• Fork StartServers number of child processes
• Wait forever and periodically,
• If children lt MinSpareServers, fork a new
  child
• If children lt MaxSpareServers, kill any child
  not
• handling request
• Child
• while (true)
• Accept client connection
• Parse request, send response
• Increment request_count
• If request_count gt MaxRequestPerChild then
  exit

11
Apache model

• Provides multiple processes to handle
  simultaneous requests
• Throttles the number of child processes
• Crashing a child process doesnt crash the
  server the parent is very stable
• Memory leaks dont take down the machine memory
  freed when child exits
• Note IIS uses similar model, but with threads
  instead of processes
• Extending Apache through modules
• Core, mime, auth, cookies, redirect

12
Apache Installation

• Download from apache.org
• http//www.apache.org/dist/httpd/binaries/solaris/
  
• gunzip c apache-tar.gz tar xvf -
• ./configure --prefix HOME/apache
• make
• make install
• /apache/bin/apachectl start
• Test by telnet localhost 8080

13
Apache Configuration (Starting)

• Port (Usually 80 when deployed must be unique
  for the machine)
• DocumentRoot (Directory for files)
• ServerName (Domain name for server)
• User (for root only specify Unix user id)
• To validate apachectl configtest

14
Apache Administration

• Starting
• apachectl start verify with ps -ef
• Stopping
• apachectl stop kills cat logs/httpd.pid
• Viewing logs
• access_log host ident authuser date request
  status bytes
• error_log date priority host error
• Log rotation (rotatelogs) excessive logging

15
Special Topics

• SSL
• What it is, how it works (handshake, key gen,
  authen, client auth)
• Traffic flows over the https port 443
• Added with mod_ssl
• Getting a server certificate (Verisign
  350-1000)
• Overhead to SSL requests

16
Special Topics

• Virtual Domain Hosting
• Serve multiple web sites from the same machine
• Name based or IP based
• Syntax
• NameVirtualHost
• ltVirtualHost gt
• ServerAdmin sinala_at_gremlin.ucsd.edu
• DocumentRoot "/home/sinala/apache/htdocs"
• ServerName gremlin.ucsd.edu
• ErrorLog logs/error_log
• CustomLog logs/access_log common
• lt/VirtualHostgt
• ltVirtualHost gt
• ServerAdmin webmaster_at_mysite.com
• DocumentRoot "/home/sinala/apache/htdocs/mysit
  e"
• Port 8080
• ServerName mysite.com

17
Special Topics

• CGI (Common Gateway Interface)
• Initial way to add interactivity
• Uses mod_cgi
• Slow new process created per request
• Perl Example /cgi-bin/rumple

18
(No Transcript)
19

• HTTP authentication
• MIME types
• Proxies
• Expected performance
• Multilanguage