Web Servers - Distributed Operating Systems - Lecture Slides, Slides of Computer Science

These are the Lecture Slides of Distributed Operating Systems which includes Neumann Bottleneck, Networked Information, Memory Hierarchy, Evidence, Latency, Communication, Intelligent Service, Communication Latency, Routing Path etc.Key important points are: Web Servers, Distributed Operating System, Problem Description, Content Providers, Huge Influxes, Larger Sites, Shutdown, Knocking Little Sites, Aim, Objectives

Typology: Slides

2012/2013

Uploaded on 03/27/2013

ekana
ekana 🇮🇳

4

(44)

370 documents

1 / 9

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
A Distributed Operating System
for Web Servers
Docsity.com
pf3
pf4
pf5
pf8
pf9

Partial preview of the text

Download Web Servers - Distributed Operating Systems - Lecture Slides and more Slides Computer Science in PDF only on Docsity!

A Distributed Operating System

for Web Servers

Problem Description

• The World Wide Web is more popular than ever and

the demands on the content providers are higher

than ever. Popular sites can now get a million of hits

a day or more.

• Sites can experience huge influxes in visits, smaller

sites can especially be affected by surges in traffic by

being mentioned by other larger sites.

• When sites experience these huge influxes they are

either forced to take down content or are effectively

shutdown.

Significance of points

  • Running a web server today requires most sites to keep extra

hardware and bandwidth on hand even though most of the time

this hardware is not needed. While it is true that internet usage is

on the raise, many sites experience momentary surges in traffic

when the larger sites mention them or they release new software.

  • Today’s website’s often incorporate server generated dynamic

content – JSP and ASP - that proxy servers have trouble caching and

even when pages do not have any it requires the proxy server to be

setup by the users or isp’s of the users rather than the server

administrators of the websites that need the help.

  • Distributed operating systems could be helpful in spreading the

load across other providers that are not currently experiencing

surges.

Significance of points continued

  • If a score-keeping service was created you could ensure that no one

was abusing the service and you could prove to the bigger sites that

they were not only taking on more of a load then they already have

  • as it is likely that they would be a source of bandwidth when a

smaller suddenly has a surge of usage.

  • The Sprite Distributed Operating System could be modified to

incorporate this service and is designed to work with powerful

individual systems, giving the owners of those systems preference

over work for remote users and to work without CPU pools and

high-speed interconnects.

  • Everyone who downloads software or surfs the web would benefit

from the pages being served from sites that are physically located

closer.

Example of web request being fulfilled with

normal web loads

www.smallsite.com Located on east coast

www.popularsite.com

Located on west coast

Surfer Located on east coast

  1. User types in URL for a popular site
  2. Popular site detects user is on east coast and points the user to a small site that is located on east coast
  3. The small site fulfills all requests from the user for a block of time or until its traffic reaches 90% capacity

Example of smaller site getting a spike in

Small east requests cost site

Popular west coast site

Popular east coast site

Smaller site is serving web pages for itself and the popular west coast site to people in the east coast

Readers of popular west coast site

Readers of the smaller site

Requests for the smaller site increase to 70% of smaller server’s capacity

Smaller site ensures that other sites with the capacity to take on additional request have complete up to date copies of it in their caches Requests for the smaller site increase to 90%

All requests for the popular west coast site are redirected to original host to be handled Requests for the smaller site increase to 95% of capacity

All new requests for pages are now handed off to the closest server until levels drop to near normal levels

Additional Readers of the small site on the west coast

Additional east cost readers of smaller site