Web Caching

1
Web Caching
2
Why Caching?

• Faster browsing experience for users
• Cache hit rate
• Traffic Prioritization
• Reduce network bandwidth requirements
  significantly
• Live media stream splitting
• Control who goes where and who does what and
  when they can do it
• Audit Employee Use of Corporate Assets
• Increase Performance, Increase Security, Improve
  Productivity, and Reduce Costs!

3
Take A Look At This Page.
4
Web Caching
Internet
Server
Lisas Desktop
5
The Cache Dilemma
?
Hit Rate
Freshness
6
Why Hit Rate is Important

• Better cache hit-rate means
• Higher effective bandwidth
• Lower avg. latency
• Improve hit-rate with
• Locality of access
• More users

7
What Content and Protocols

• HTTP 1.0 Basic protocol
• Send Request based on fix number of verbs
• GET
• HEAD
• POST
• Receive response, meta-data, content

8
What Content and Protocols

• Example
• GET /pub/www/index.html HTTP/1.0
• Response
• HTTP/1.1 200 OK
• Server Microsoft-IIS/5.0
• Date Sat, 19 Oct 2002 054653 GMT
• Expires Sun, 20 Oct 2002 160000 GMT
• Content-Length 2291
• Content-Type text/html
• Cache-control private

9
What Content and Protocols

• Example if-modified-since
• GET /pub/www/index.html HTTP/1.0
• If-Modified-Since Sat, 19 Oct 2002 194331 GMT
• Response
• HTTP/1.1 200 OK
• Server Microsoft-IIS/5.0
• Date Thu, 13 Jul 2000 054653 GMT
• Expires Sun, 20 Oct 2002 160000 GMT
• Content-Length 2291
• Content-Type text/html
• Cache-control private

10
What Content and Protocols

• Example if-modified-since
• GET /pub/www/index.html HTTP/1.0
• If-Modified-Since Sat, 19 Oct 2002 194331 GMT
• Response
• HTTP/1.1 304 Not Modified

11
Basic caching algorithm

• Pages may be
• Fresh up-to-date
• Expired current date gt expiration date
• Stale old

12
Basic caching algorithm - 2

• If (page is in the cache)
• if ( page is expired or stale )
• Get from server - if-modified-since
• If not modified, Get from cache
• else Get from Server
• else
• Get from cache
• Else
• Get from Server

13
Basic caching algorithm - 3

• If cache has space
• Store the file
• Else
• Delete expired from cache
• Delete stale from cache
• Delete LRU from cache
• Delete largest/smallest from cache?
• Store the file

14
Proxy Details

• Without Proxy

GET / HTTP/1.1 Host localhost1235 User-Agent
Mozilla/5.0 (Macintosh U Intel Mac OS X en-US
rv1.8.0.7) Accept image/png,/q0.5 Accept-Lan
guage en-us,enq0.5 Accept-Encoding
gzip,deflate Accept-Charset ISO-8859-1,utf-8q0.
7,q0.7 Keep-Alive 300 Connection keep-alive
15
Proxy Details

• With Proxy

GET http//star.cs.byu.edu/CS360 HTTP/1.1 Host
star.cs.byu.edu User-Agent Mozilla/5.0
(Macintosh U Intel Mac OS X en-US
rv1.8.0.7) Accept text/xml,application/xml,text/
plainq0.8,image/png,/q0.5 Accept-Language
en-us,enq0.5 Accept-Encoding
gzip,deflate Accept-Charset ISO-8859-1,utf-8q0.
7,q0.7 Keep-Alive 300 Proxy-Connection
keep-alive
16
Zipfs Law

• In a corpus of natural language utterances, the
  frequency of any word is roughly inversely
  proportional to its rank in the frequency table.
• The most frequent word will occur approximately
  twice as often as the second most frequent word,
  etc.
• Example In the Brown Corpus, the is the most
  frequently occurring word and accounts for nearly
  7 of all word occurrences. (69971 of slightly
  over 1 million)
• 2nd place of - slightly over 3.5 (36411)
• 3rd place and (28852)
• Only 135 words are needed to account for half of
  the Brown Corpus

17
Zipfs law

• Zipfs law The frequency of an event P as a
  function of rank i is a power law function
• Pi ? / ia where a 1

18
Zipfs law

• Observed to be true for
• Frequency of written words in English texts
• Population of cities
• Income of a company as a function of rank

19
Zipfs law and web access

• For a given server, page access by rank follows
  Zipfs law
• Web requests from a fixed population of users
  follows Zipfs law 0.64 lt a lt 0.83

20
Observations

• Top 1 of all documents account for 20 - 35 of
  proxy requests
• Top 10 account for 45 - 55 of requests
• It takes 25 to 40 of all documents to account
  for 70 of requests
• It takes 70 to 80 of all documents to account
  for 90 of requests

21
Observations
22
Observations

• For an infinite sized cache, the hit-ratio for a
  web-proxy grows in a log-like fashion as a
  function of the client population of the proxy
  and the number of requests seen by the proxy.

23
Local URL Resolution Protocol

• Peer-to-Peer web-cache
• Bootstrapping Peer Discovery
• UDP broadcast
• Content Location
• UDP broadcast for content
• Content Delivery
• Direct Download from single peer

24
(No Transcript)