2023-09-14

Introduction

Summary

keywords

TODO

HW performance calculation on installing web cache

Exercise*

Next time HOL blocking

---

Recap

What kind of Application need which transport protocol?

TCP & UDP have no encryptions, throughput guarantee, timing,,

Web and HTTP

HTTP uses TCP TCP is closed after a certain amount of time, or certain amount of transmissions HTTP is stateless. headers do not store any stateful meanings.

Non-persistent TCP no longer used. High RTT One single connection on one transmssion

---

request and response

all human readable. Written in ASCII codes)

HTTP request message

composed of * request line (method, url, version) * header lines (header values) * body (data)

in the request message,

• GET, POST, HEAD, PUT request types

• lines are broken down by \r

For normal GET request, there's no body

GET

• send data to server by URL query (after question mark).

POST

• form input

HEAD

• request headers only.

• only used when checking server up status.

• will only send essential info

PUT

• upload a file; alternative for FTP.

• completely replaces file that exist at specified URL

Response message

composed of

• status line (HTTP/1.1 200 OK)

• header lines

• body

in the header * Date, Server, Last-Modified, * Etag, * Accept-Ranges, * (Content-Length, Content-Type..) * and more..

Etag is used for caching page purposes. version controlling for page. ex . ( Etag : "abb-43kfnnldocmmdkvndls")

#todo : what is Accept-Ranges

Response status codes

code	name	description
200	OK	success
301	Moved Permanently	request object moved, new location specified
400	Bad request	not understood by server
404	Not found	not found
505	HTTP Version Not supported	-

and more..

Cookies

client-side tool to make WEB stateful.

• in HTTP, all request are independent of each other

• browser stores the cookie

• widely used in server identifying a client browser&machine

  • Authorization, shopping carts, recommendations, user session state(e-mail..)

• when written in browser, it transcends time and can be used in the long apart future and show the seamless cookie-specific action.

four components of data storages

• cookie header line ( of response message)

• cookie header in next HTTP request message)

• cookie file kept on user's host, managed by user's

• back-end database at web site

#todo : check the python example on the slides.

import requests

response = requests.get("https://www.google.com")

if response.status_code == 200:
	cookies = response.cookies
	for cookie in cookies:
		print(f"Cookie Name: {cookie.name}")
		print(f"Cookie value: {cookie.value}")

else :
	print("HTTP Error:",response.status_code,response.reason)

Cookie Name: 1P_JAR
Cookie value: 2023-09-14-04
Cookie Name: AEC
Cookie value: Ad49MVFXS3NX5m3l8PO4X5FUaFZvQI8b1Czl9-Ym1cAxZn7fTXSC5fXdAw
Cookie Name: NID
Cookie value: 511=m0D8Dti7MEGMgrBAZI_1a3ZfZ8xr9ydTVeHn60MicQtJztyPIo5Qj7U73zEnhAm8i1KeRJfAjsANOwpxvqaKfemnBYIWKnfl_OZOjFbDw6rMPjH_aEG9PFVAp20YMu8ZOpIaj7b7NBSauxhKPq3-ceyC3o-pyKj-X-5_T1G0ou4

• Cookie has trade-offs on privacy

  • it is a info a lot about you.

• How to keep state? (other than cookies?)*

  • cookies inside the HTTP message

  • protocol endpoint ?

  • #todo what is protocol endpoint

Web Caches

Intermediate servers that saves a server and sending pages instead of the origin server

• satisfy client request without involving origin server

• Act as a client against origin server, act as a server for end-users

• server tells cache about whether the object can be cached or not

  • Cache-Control: max-age=<seconds>

advantage

• makes less load on the origin server

• reduce response time

• reduces traffic on the backbone network

following is a small server

from flask import Flask, make_response

app = Flask(__name__)

@app.route('/')
def hello():
	response = make_response("hello, world")
	respon.headers["Cache-Control"]= 'public, amx-age=3600'
	return response

if __name__ == '__main__':
	app.run()

Caching Scenario

![[../images/20230914140127.png]]

$$L = 100K\;bits\; a = 15/sec\; R = 1.54 MBps\; access link utilization = La/R = .97 LAN utilization = 0.0015 delay : 2sec + access link delay(minutes) + LAN delay$$

So slow. How will it be 해결?

1. buy a faster link high R cost-ineffective

2. install a web cache #todo : make performance calculations

Conditional GET

Don't send object if cache has up-to-date cached version

• no object transmission delay

Client sends a If-modified-since header message of cache that they have304 Not modified status code tells us "you don't need a newer one"

![[../images/20230914140912.png]]