caching web contents on a long latency and often

Transcrição

caching web contents on a long latency and often
CACHING WEB CONTENTS
ON A LONG LATENCY AND
OFTEN BROKEN LINES
João Paulo Barrocas | Universidade Lusófona
INTRODUCTION
• Often broken communication lines
– Remote Locations
– Natural catastrophes
– Satellite access
• Latency
– Outer space communications
INTRODUCTION
• Web access problems
– Often broken communication lines
• Browser’s implicit implies reloading
• Long time transactions
– Latency
• TCP no suitable
• Browser’s implicit implies reloading
• Long time transactions
PROPOSED SYSTEM
Latency
BACKBONE
PROXY A
PROXY B
BACKBONE
Often Fail
• Isolate problematic lines
• Dependent from proxy
– Cache ~30 % of Web Content (static HTML)
– Some specifically cases of dynamic pages (php, asp, cgi, etc)
• Limitations
– Real time applications
– Proxies' bi-directionality
PROPOSED SYSTEM
WEB Gateway
Proxy I
Proxy O
Web Server
DB
Communication Layer
Backbone
Latency
Often Fail
PROPOSED SYSTEM
• Limitations
– Depends on proxies capacities
• New solutions
– A two-way, full-duplex web caching
– Implementation of new facilities is easier and proxy’s independent
– No need to change used proxies
PROPOSED SYSTEM
• Proxies Limitations
– Minor percentage of dynamic contents
• How to
– Dynamic Web Page caching.
PROCEDIMENTO
• Liste todos os passos utilizados para a realização da experiência.
• Não se esqueça de numerar os passos.
• Adicione fotografias às experiências.
DYNAMIC WEB PAGE CACHING
Web Server 1
A. Send all elements to WG A
can replicate all site. (Data
included)
B. Send posterior updates on files and
data to WG A can steal replicate all site.
WEB
Web Gateway A
C. Send all elements to WG
B can replicate all site.
(Data included)
D. Send posterior updates on
files and data to WG A can steal
replicate all site.
Web Gateway B
End User
E. Ask for some
intensive operation, or
notify updates on data.
F. Return the result of
the operation
requested on E.
CONCLUSIONS
• To Test
– Additional Traffic impact
– Overload off the new process’s
• To think
– How information will flow between Gateways?
– How will we construct this model without being to much intrusive?
REFERENCES
• S. Burleigh. Licklider Transmission Protocol - Motivation. RFC-5325.
IEFT - 2008
• JPL – Jet Propulsion Laboratory – http://www.jpl.nasa.gov