thesis.PDF

Similar documents
thesis-shk

1217 WebTrafMon II

DBPIA-NURIMEDIA

°í¼®ÁÖ Ãâ·Â

PowerPoint 프레젠테이션

Intra_DW_Ch4.PDF

SMB_ICMP_UDP(huichang).PDF

Voice Portal using Oracle 9i AS Wireless

<C0CCBCBCBFB52DC1A4B4EBBFF82DBCAEBBE7B3EDB9AE2D D382E687770>

[ReadyToCameral]RUF¹öÆÛ(CSTA02-29).hwp

歯최덕재.PDF

SLA QoS

DBPIA-NURIMEDIA

Remote UI Guide

歯홍원기.PDF

학습영역의 Taxonomy에 기초한 CD-ROM Title의 효과분석

#Ȳ¿ë¼®

Cache_cny.ppt [읽기 전용]

µðÇÃÇ¥Áö±¤°í´Ü¸é

High Resolution Disparity Map Generation Using TOF Depth Camera In this paper, we propose a high-resolution disparity map generation method using a lo


¼Ł¿ï¸ðµåÃÖÁ¾

ÀÌÀç¿ë Ãâ·Â

Software Requirrment Analysis를 위한 정보 검색 기술의 응용

TTA Verified : HomeGateway :, : (NEtwork Testing Team)

THE JOURNAL OF KOREAN INSTITUTE OF ELECTROMAGNETIC ENGINEERING AND SCIENCE. vol. 29, no. 10, Oct ,,. 0.5 %.., cm mm FR4 (ε r =4.4)

3ÆÄÆ®-14

À±½Â¿í Ãâ·Â

歯1.PDF

µðÇÃÇ¥Áö±¤°í´Ü¸é

untitled

VOL /2 Technical SmartPlant Materials - Document Management SmartPlant Materials에서 기본적인 Document를 관리하고자 할 때 필요한 세팅, 파일 업로드 방법 그리고 Path Type인 Ph

SRC PLUS 제어기 MANUAL

PowerChute Personal Edition v3.1.0 에이전트 사용 설명서

thesis

ORANGE FOR ORACLE V4.0 INSTALLATION GUIDE (Online Upgrade) ORANGE CONFIGURATION ADMIN O

09È«¼®¿µ 5~152s

thesis

인문사회과학기술융합학회

Mstage.PDF

ARMBOOT 1

I

04-다시_고속철도61~80p

V28.

Interstage5 SOAP서비스 설정 가이드

0. 들어가기 전

6.24-9년 6월

Intro to Servlet, EJB, JSP, WS

<35335FBCDBC7D1C1A42DB8E2B8AEBDBAC5CDC0C720C0FCB1E2C0FB20C6AFBCBA20BAD0BCAE2E687770>

05( ) CPLV12-04.hwp

UDP Flooding Attack 공격과 방어

I T C o t e n s P r o v i d e r h t t p : / / w w w. h a n b i t b o o k. c o. k r



PBNM CIM(Common Information Model) DEN, COPS LDAP 21 CIM (Common Information Model) CIM, specification schema [7]

< C6AFC1FD28B1C7C7F5C1DF292E687770>

API STORE 키발급및 API 사용가이드 Document Information 문서명 : API STORE 언어별 Client 사용가이드작성자 : 작성일 : 업무영역 : 버전 : 1 st Draft. 서브시스템 : 문서번호 : 단계 : Docum

untitled

44-4대지.07이영희532~

Analyst Briefing

bn2019_2

VZ94-한글매뉴얼

Journal of Educational Innovation Research 2019, Vol. 29, No. 1, pp DOI: * Suggestions of Ways

(Exposure) Exposure (Exposure Assesment) EMF Unknown to mechanism Health Effect (Effect) Unknown to mechanism Behavior pattern (Micro- Environment) Re

SchoolNet튜토리얼.PDF

표현의 자유

歯이시홍).PDF

03.Agile.key

DBPIA-NURIMEDIA

Secure Programming Lecture1 : Introduction

PowerPoint 프레젠테이션

03이경미(237~248)ok

untitled

Smart Power Scope Release Informations.pages

PWR PWR HDD HDD USB USB Quick Network Setup Guide xdsl/cable Modem PC DVR 1~3 1.. DVR DVR IP xdsl Cable xdsl Cable PC PC DDNS (

untitled


USB USB DV25 DV25 REC SRN-475S REC SRN-475S LAN POWER LAN POWER Quick Network Setup Guide xdsl/cable Modem PC DVR 1~3 1.. DVR DVR IP xdsl Cable xdsl C

0125_ 워크샵 발표자료_완성.key

PCServerMgmt7

09권오설_ok.hwp

The Self-Managing Database : Automatic Health Monitoring and Alerting

Web Application Hosting in the AWS Cloud Contents 개요 가용성과 확장성이 높은 웹 호스팅은 복잡하고 비용이 많이 드는 사업이 될 수 있습니다. 전통적인 웹 확장 아키텍처는 높은 수준의 안정성을 보장하기 위해 복잡한 솔루션으로 구현

4 CD Construct Special Model VI 2 nd Order Model VI 2 Note: Hands-on 1, 2 RC 1 RLC mass-spring-damper 2 2 ζ ω n (rad/sec) 2 ( ζ < 1), 1 (ζ = 1), ( ) 1

김기남_ATDC2016_160620_[키노트].key

45-51 ¹Ú¼ø¸¸

1 : UHD (Heekwang Kim et al.: Segment Scheduling Scheme for Efficient Bandwidth Utilization of UHD Contents Streaming in Wireless Environment) (Specia

<313120C0AFC0FCC0DA5FBECBB0EDB8AEC1F2C0BB5FC0CCBFEBC7D15FB1E8C0BAC5C25FBCF6C1A42E687770>

The characteristic analysis of winners and losers in curling: Focused on shot type, shot accuracy, blank end and average score SungGeon Park 1 & Soowo

DBPIA-NURIMEDIA

ISO17025.PDF

Microsoft PowerPoint - XP Style

THE JOURNAL OF KOREAN INSTITUTE OF ELECTROMAGNETIC ENGINEERING AND SCIENCE Feb.; 29(2), IS

歯A1.1함진호.ppt

Subnet Address Internet Network G Network Network class B networ

±è¼ºÃ¶ Ãâ·Â-1

슬라이드 제목 없음

ETL_project_best_practice1.ppt

Backup Exec

Transcription:

1 1 2 4 21 4 22 7 221 7 222 9 223 9 3 11 31 12 32 12 33 (Prefetching Parameters)13 34 (Web Traffic Trace)14 35 (Prefetching Time)18 36 (Performance Metrics)19 4 21 41 Prefetchable Object List Generator 22 i

411 HTTP/10 Response Header 24 412 HTTP/11 Request Header 27 413 Squid 29 414 Freshness 32 42 Request Generator 36 5 37 51 37 52 38 53 39 6 41 61 Accuracy41 62 Wasted Bandwidth46 63 Request Saving 47 64 Bandwidth Saving 49 65 Summary51 7 55 [ ] 57 ii

1 1 2 2 3 16 4 17 5 17 6 17 7 18 8 21 9 Prefetchable Object List 23 10 HTTP/10 Response Header 24 11 Cache-Control General Header Field 28 12 30 13 31 14 STALE 32 15 Access Log Format 33 16 Store Log Format 34 17 35 18 40 19 42 iii

20 43 21 44 22 44 23 45 24 Off-peak Periods 47 25 48 26 50 27 Peak Periods 51 iv

1 Cache Summary Statistics 15 2 15 3 16 4 42 5 43 6 45 7 46 8 Off-peak Periods 47 9 48 10 Cache Byte Hit Ratio with and without Prefetching 49 11 Peak Periods 50 12 Performance Mateics 51 13 Peak Periods Off-peak Periods 52 14 Peak Periods 53 v

vi

MCC 9851M08, Sook-Hyang Kim, A Statistical, Batch and Proxy-Initiated Web Prefetching Scheme for Efficient Internet Bandwidth Usage,,, 2000, 60P, Advisor: J Won-Ki Hong, Text in Korean ABSTACT As the number of World-Wide Web (Web) users grows, Web traffic continues to increase at an exponential rate Currently, Web traffic is one of the major components of Internet traffic Also, high bandwidth usage due to Web traffic is observed during peak periods while leaving bandwidth usage idle during off-peak periods One of the solutions to reduce Web traffic and speed up Web access is through the use of Web caching Unfortunately, Web caching has limitations for reducing network bandwidth usage during peak periods In this thesis, we focus our attention on the use of a prefetching algorithm for reducing bandwidth during peak periods by using off-peak period bandwidth We propose a statistical, batch, proxy-side prefetching algorithm that improves cache hit rate while only requiring a small amount of storage We present simulation results based on Web proxy trace and show that this prefetching algorithm can reduce peak time bandwidth using off-peak bandwidth

1 World Wide Web( ),,, [20, 21], (Internet traffic) (Web traffic) (bottleneck) [25, 26] (bandwidth) peak periods, off-peak periods 1 16 subnet 2 (1999 10 15 1999 10 28 ), 2 1 1

2 (156624826 byte) peak periods off-peak periods peak periods 14:00 16:00 18:00 04:00, 12 off-peak periods 04:00 14:00 16:00 18:00 12 1 2 peak periods off-peak periods peak periods off-peak periods, (Web caching) [10], [11, 12], 2

[7, 27, 28] peak peiods off-peak periods, off-peak periods peak periods [4, 5, 11] (caching server) (Web prefethcing) peak periods off-peak periods,, off-peak periods peak periods 2 3 4 5, 6 7 3

4 2 21,, Padmanabhan Mogul [2] (prediction) (prediction algorithm) Griffioen Appletion[13],

hyperlink dependency graph graph A B A B arc (edge) arc A B dependecy graph (Web server traces) trace-driven 45% 2 Wang Crowcroft HotList Manager delay [1] deterministic client-initiated prefetching, Deterministic prefetching Bestavros Server-initiated prefetching [14] D i, D j Bestavros D i time interval T w D j p[i,j] advice Bestavros trace-driven 10% 23% cache miss rate 5

Kroeger, 60% [15] Chinen Yamaguchi Wcol [3, 17] HTML (parsing) - (pre-pushing) Wcol 246 167 281 Jacobson Cao - [18] - (push) (Web access traces) (overhead) (accuracy) - (pre-push scheme) - 10% 18%, (request) 12% Makatos 6

Chronak Top-10 Approach [16, 24] client-proxy-server framework Top-10 (Web server trace), 10% 40%,,, 22,, 221,, 7

Server-initiated prefetching :, hyperlink (pushing) [2, 14, 15], Client-initiated prefetching : (agent) [1, 2, 13] Client-initiated prefetching (Web access pattern) Proxy-initiated prefetching :, [3, 17, 18] 8

222 (statistical prefetching)[5, 8, 9] (deterministic prefetching) [1] Statistical prefetching : (access log) Deterministic prefetching :, 223 (response time) 9

Prefetching for Response Time Reduction :, [2, 5, 13, 16, 18] peak periods Prefetching for Balanced Bandwidth Usage : peak periods off-peak periods Batch prefetching balanced bandwidth usage 10

3 Server-initiated Prefetching Client-initiated Prefetching peak bandwidth usage, (parameter) real-world traffic pattern 16 (subnet) (performance metrics) 11

31??,???? 32?? ( :, 1 )??, (accuracy) prefetchable objects 12

(expiration time) (cache miss)?? 33 (Prefetching Parameters),,,,??,?? Traditional prefetching scheme n 13

?? 100M?? off-peak periods peak periods off-peak periods off-peak periods off-peak periods 35 34 (Web Traffic Trace) real-world traffic pattern 16 (subnet) 1999 10 15 ( ) 10 28 ( ) 2 2 447 (cache object hit rate) 5496% (cache byte hit rate) 3142% (request) 4,077,308 total bytes 502G 291236 bytes 36G 1 14

1 Cache Summary Statistics Number of Client Making Requests 447 Cache Object Hit Rate 5496 % Cache Byte Hit Rate 3142 % Total Number of Objects Requested 4,077,308 Total Bytes to Clients 502 Gbyte 2 (reference count) 7075% 2 2925% 121,883 3 2 (%) 1 86,227 7075 2 16,690 1369 3 6,462 53 4 3,451 283 5 9,053 743 Total 121,883 100 15

3 3 3 total size 1 6812%, 2 3188% total bytes 217227 Mbyte 4 3 (Mbyte) (%) 1 147975 6812 2 31079 1431 3 11820 544 4 7401 341 5 18952 872 Total 217227 100 16

4 5 6 total bytes 5 6 17

35 (Prefetching Time) off-peak periods peak periods off-peak periods off-peak periods off-peak periods off-peak periods off-peak periods off-peak periods 80% (125299861 byte) off-peak peirods off-peak periods 04:00 13:00, peakperiods 13:00 04:00 off-peak periods off-peak periods (04:00 13:00) prefetchable object 7 80% 7 18

36 (Performance Metrics),?? Request Saving : hit Request saving?? Bandwidth Saving : Bandwidth saving peak periods bandwidth usage peak periods peak periods?? Accuracy : accuracy (prefetched object hit rate) (prefetched byte hit rate) total bytes total bytes Accuracy?? Wasted Bandwidth : 19

offpeak periods off-peak periods 20

4 8 Prefetching system Caching Server prefetchable object list Prefetchable Object List Generator accesslog storelog cache Request Generator Caching Server clients request response request response request response Internet 8, Squid[7] Squid freeware, Cobalt Network CacheRaQ[29] Packetstorm Technologies Webspeed[30] Squid Squid 21

Squid transparent Transparent (browser) (data path) [31] configuration default gateway Prefetchable Object List Generator Request Generator (Squid) Squid, Squid request access log, store log squid (accesslog, storelog) Prefetchable Object List Generator prefetchable object list Request Generator off-peak periods prefetchable object list (request) Squid Squid add-on 41 Prefetchable Object List Generator Prefetchable Object List Generator 22

Squid access log store log access log Squid requests store log Prefetch Object List Generator Squid (refresh algorithm) prefetchable object HTTP response message Squid (3 days) [7] 9 prefetchable object list 9 Prefetchable Object List Prefetchable Object List Generator Squid, store log store log HTTP response header HTTP response header HTTP request header Squid 23

411 HTTP/10 Response Header HTTP [22, 23, 33] / (Request/Response), ( : GET, HEAD, POST) TCP,, TCP HTTP method, URI, protocol version,,,,,, HTTP response header HTTP/10 response header 10 Full-Response = Status-Line *(General-Header Response-Header Entity-Header) CRLF [Entity-Body] Status-Line = HTTP-Version Status-Code Reason-Phrase CRLF General-Header = Date Progma Response-Header = Location Server WWW-Authenticate Entity-Header = Allow Content-Encoding Content-Length Content-type Expires Last-Modified extension-header extension-header = HTTP-header 10 HTTP/10 Response Header 24

Full-Response Status-Line HTTP (Status-Code) (Reason-Phrase) CRLF Entity CRLF Status-Code?? 1xx : Informational -?? 2xx : Success -?? 3xx : Redirection -?? 4xx : Client Error -?? 5xx : Server Error - (response message), General-Header, Request-Header, Response-Header, Entity-Header RFC 822 31 [32] 25

General Header Full-Request Full-Response, General Header Date Progma Date Progma / Response Header Status-Line Request-URI Entity-Header Entity-Body,,, (metainformation), entity body Squid HTTP response header?? Date : General Header Date, RFC 822[32] orig-date Date : Wed, 15 Oct 1999 08:12:31 GMT?? Expires : Entity Header Expires, 26

Expires Expires : Mon, 01 Nov 1999 16:00:00 GMT?? Last-Modified : Entity Header Last-Modified Last-Modified Last-Modified : Tue, 14 Oct 1999 12:45:26 GMT 412 HTTP/11 Request Header HTTP/10, URL TCP / / congestion congestion 27

, HTTP/11 [23] HTTP/11, Squid Cache-Control Request Header max-age HTTP/10 HTTP HTTP/11, HTTP/11 General Header Cache-Control Cache-Control (directive) (cache directive) Cache-Control 11 11 cache-response-diective Cache-Control Cache-Control = "Cache-Control" ":" 1#cache-directive cache-directive = cache-request-directive cache-response-directive cache-request-directive = "no-cache" "no-store "max-age" "=" delta-seconds "max-stale" [ "=" delta-seconds ] "min-fresh" "=" delta-seconds "no-transform" "only-if-cached" cache-extension cache-extension = token [ "=" ( token quoted-string ) ] 11 Cache -Control General Header Field 28

cache-request-directive max-age Squid (maximum object age) max-age, max-age max-age Squid CLIENT_MAX_AGE 413 Squid, (expiration time) (cache miss) Squid 13 Squid default value OBJ_DATE OBJ_LASTMOD Expires OBJ_DATE, OBJ_LASTMOD, Expires HTTP Respons Header [22] CLIENT_MAX_AGE HTTP/11 Cache-Control Request Header 29

[23] Squid refresh_pattern CONF_MIN, CONF_PERCENT, CONF_MAX squidconf HTTP Response Header Expires default Expires [7] NOW, OBJ_AGE LM_AGE LM_FACTOR OBJ_AGE LM_AGE 12 OBJ_DATE, Expires, OBJ_LASTMOD millisecond UNIX time ( : Fri Oct 15 04:00:27 1999 GMT = 939927627725) CLIENT_MAX_AGE, OBJ_AGE, LM_AGE, LM_FACTOR second CONF_MAX CONF_MIN minute CONF_PERCENT % OBJ_DATE Expires OBJ_LASTMOD : : : CLIENT_MAX_AGE : OBJ_AGE = NOW - OBJ_DATE (sec) LM_AGE = OBJ_DATE - OBJ_LASTMOD (sec) LM_FACTOR = OBJ_AGE / LM_AGE (sec) CONF_MAX CONF_MIN : 4320 (min, 3 days) : 0 (min) CONF_PERCENT : 02 (20%) 12 30

Is CLIENT_MAX_AGE present in the request? yes Is OBJ_AGE more than CLIENT_MAX_AGE? yes STALE(1) no no Is Expires present in the response? yes Has the object already expired? yes no STALE(2) FRESH no Is OBJ_AGE more than CONF_MAX? yes STALE(3) no Is OBJ_DATE more than OBJ_LASTMOD? yes LM_FACTOR less than the CONF_PERCENT? yes no FRESH STALE(4) no Is OBJ_AGE less than CONF_MIN? yes no FRESH STALE(5) 13 14 STALE 5 Expires, OBJ_DATE, OBJ_LASTMOD, NOW HTTP CLIENT_MAX_AGE 100 sec OBJ_AGE 120 sec OBJ_AGE CLIENT_MAX_AGE STALE (STALE(1)) Expires NOW, (STALE(2)) OBJ_AGE CONF_MAX (STALE(3)) 3 OBJ_DATE OBJ_LASTMOD LM_FACTOR (30%) CONF_PERCENT (20%) (STALE(4)) 31

OBJ_AGE (5760 min) CONF_MIN (0 min) STALE (STALE(5)) STALE(1) CLIENT_MAX_AGE : 100 (sec) OBJ_AGE : 120 (sec) STALE(2) EXPIRES : Fri, 01 Oct 1999 16:00:00 GMT NOW : Sun, 03 Oct 1999 12:01:20 GMT STALE(3) OBJ_AGE : 5760 min (4 days) CONF_MAX :4320 min (3 days) STALE(4) OBJ_LASTMOD : Wed, 29 Sep 1999 16:00:00 GMT OBJ_DATE : Fri, 01 Oct 1999 16:00:00 GMT LM_FACTOR : 30 % CONF_PERCENT : 20 % STALE(5) OBJ_AGE : 5760 min (4 days) CONF_MIN : 0 min 14 STALE 414 Freshness FRESH, STALE STALE squid OBJ_DATE, 32

OBJ_LASTMOD, Expires squid storelog, CONF_MAX, CONF_MIN, CONF_PERCENT squid squidconf CLIENT_MAX_AGE 15 Squid access log, 16 Squid store log 15 Access Log Format?? Time UNIX time milliseconds?? Elapsed (connection)?? Remotehost IP?? Code/Status Code ( : TCP_HIT, TCP_MISS, etc) Status HTTP status code?? Byte?? Method HTTP request Method ( : GET, HEAD, POST)?? URL URL 33

Time Action Status OBJ_DATE OBJ_LASTMOD Expires Type Len Method URL 16 Store Log Format store log Time, Status, Method, URL access log OBJ_DATE, OBJ_LASTMOD, Expires 413?? Action RELASE, SWPIN, SWPOUT RELASE SWAPOUT SWAPIN?? Type, text/html, image/gif?? Len expect-len real-len, expect-len HTTP Content-Length Response Header, real-len NOW, fresh stale access log stale stale Squid NOW 34

NOW? Current Pr efetchngtime? PrefetchingFrequency Current PrefetchingTime : Pr efetchingfrequency : off-peak periods (prefetching frequency) 17 10 15 04 16 04, 17 04 (16 04:00) 10 15 04:00 10 16 04:00 access log store log Last prefetching time Currnt Prefetching time Next prefetching time Input Data (logs) Prefetching Frequency 10/15 4:00 10/16 4:00 10/17 4:00 17 35

42 Request Generator Request generator Prefetchable Obejct List off-peak periods Prefetchable Object List Generator Prefetchable Object List request off-peak periods request HTTP command-line Web client wget [19] crontab off-peak periods Request Generator Off-peak periods off-peak WAN SNMP agent SNMP agent (polling), 36

5 (input parameter) 51 trace-driven CacheRaQ[29] 16 (subnet) Squid (input parameter)?? Logs : 1999 10 15 10 28 2 access log store log?? Off-peak periods : 35 off-peak periods 04:00 13:00 04:00 37

?? : 33 52?? : Squid, 413 NOW 414 52 (Web traffic trace),, (Prefetchable Web Objects) (expiration time) 1, 2, 3, 4, 5 38

off-peak periods 53 18 Prefetchable Object List Generator off-peak periods prefetchable object list performance analyzer prefetchable object list request saving bandwidth saving accracy wasted bandwidth Prefetchable Object List Generator access log store log prefetchable object list 39

Simulation Model accesslog storelog Prefetchable Object List Generator Prefetchable Object List Performance Analyzer Performance Metrics With Prefetching Performance Metrics Without Prefetching 18 Performance Analyzer prefetchable object list access log prefetchable object list access log request saving ( ), bandwidth saving ( ) accuracy (, ) access log wasted bandwidth request saving bandwidth saving accuracy wasted bandwidth 40

6 3 accuracy wasted bandwidth, performance parameter request saving bandwidth saving 61 Accuracy Accuracy 4 10 15 10 28 daily prefetchable object list 10 16 10 29 1 1522% 5 8714% 1 prefetchable object 5064% 1955% linear 41

4 (%) (%) 1 16,274 (5064 %) 2,476 (1955 %) 1522 % 2 5,525 ( 172 %) 2,286 (1805 %) 4138 % 3 3,008 ( 936 %) 1,804 (1424 %) 5996 % 4 1,966 ( 612 %) 1,430 (1129 %) 7276 % 5 5,358 (1668 %) 4,469 (3687 %) 8714 % Total 32,131 (10000 %) 12,665 (10000 %) 3942 % 19 prefetched object hit 20 19 42

20 5 1 868% 5 7264% accuracy 5 (%) (%) 1 545,399,428 (6343%) 47,336,276 (2502%) 868 % 2 154,356,860 (1795%) 56,170,230 ( 297%) 3639 % 3 64,124,072 ( 746%) 25,570,556 (1352%) 3989 % 4 33,972,272 ( 395%) 15,026,692 ( 794%) 4423 % 5 62,024,975 ( 721%) 45,057,416 (2382%) 7264 % Total 859,877,607 (100%) 189,161,170 (100%) 22 % 21 43

22 21 22 6 1, 2, 3, 4 5 44

2 6426% 1 3942%25% 2 461% 1 22% 23 6 1 394 % 22 % 2 643 % 451% 3 765 % 535 % 4 833 % 626 % 5 871 % 726 % 23 45

62 Wasted Bandwidth Wasted bandwidth 7 1 67072 Mbyte (3775043 Mbyte) 178% 2 17265 Mbyte 46% 2 1 4 7 1 67072 M 178 % 2 17265 M 46 % 3 7447 M 20 % 4 3591 M 10 % 5 1697 M 05 % off-peak periods off-peak periods off-peak 796356 Mbyte 1 off-peak 519%, off-peak periods 2 283%, 3 167%, 4 108%, 5 72% 8 off-peak periods 24 46

8 Off-peak Periods Off-peak periods 1 519 % 2 283 % 3 167 % 4 108 % 5 72 % 24 Off-peak Periods 63 Request Saving 9 55%, 1 47

43% 593% 2 353 %, 3 272%, 4 21%, 5 161% 1 2 077% 9 ( ) 55 % ( + ) ( 1 ) 593 % ( + ) ( 2 ) 585 % ( + ) ( 3 ) 577 % ( + ) ( 4 ) 571 % ( + ) ( 5 ) 567 % 25 25 48

64 Bandwidth Saving 10 314%, 1 501%, 2 378 %, 3 227%, 4 159%, 5 119% 1 2 123% 10 ( ) 314 % ( + ) ( 1 ) 364 % ( + ) ( 2 ) 352 % ( + ) ( 3 ) 337 % ( + ) ( 4 ) 330 % ( + ) ( 5 ) 326 % 26 Request saving 49

26 peak periods 11 peak periods 27 11 Peak Periods Peak periods 1 64 % 2 48 % 3 29 % 4 20 % 5 15 % 50

27 Peak Periods 65 Summary 12 Request saving 12 Performance Mateics Bandwidth saving Wasted bandwidth Prefetched object hit rate Accuracy Prefetched byte hit rate 1 430 % 501 % 178 % 394 % 22 % 2 353 % 378 % 46 % 643 % 451% 3 272 % 227 % 20 % 765 % 535 % 4 210 % 159 % 10 % 833 % 626 % 5 162 % 119 % 05 % 871 % 726 % off-peak periods peak periods 51

off-peak periods peak periods 13 13 Peak Periods Off-peak Periods Peak periods off-peak periods 1 64 % 519 % 2 48 % 283 % 3 29 % 167 % 4 20 % 108 % 5 15 % 72 % E p? B s / B w E B B s p w : : ( Mbyte) : ( Mbyte) 1, 2, 3, 4, 5, 52

peak periods peak periods peak periods peak periods 4% peak periods 14 14 Peak Periods B s (Mbyte) B w (Mbyte) E p (Mbyte) Peak periods 1 18916 67072 028 64 % 2 14182 17265 082 48 % 3 8565 7447 115 29 % 4 6008 3591 167 20 % 5 4506 1697 266 15 % 14 peak periods 4% 1, 2 2 082 1 028 2 peak periods 4% 2 53

2 141,824,895 bytes peak periods 54,000 seconds (19 hours) peak periods peak periods = (141824895*8)/54000 = 2052 Kbps 2 peak periods 2052 Kbps 54

7 peak periods, off-peak periods peak periods request saving 353% bandwidth saving 378% 2 6426%, 451% peak bandwidth usage 2052 Kbps peak periods prefetching scheme squid add-on feature 55

Web traffic trace off-peak periods (eg, 20:00~08:00),, real-time prefetchable object list 56

[1] Z Wang and J Crowcroft, Prefetching in World Wide Web, IEEE Globecom 96, http://wwwcsuclacuk/staff/zwang/papers/prefetchpsz [2] V Padmanabhan and J Mogul, Using Predictive Prefetching to Improve World Wide Web Latency, Computer Communication Review, 26(3):22-36, July 1996 [3] Ken-ichi Chinen and Suguru Yanaguchi, An Interactive Prefetching Proxy Server for Improvement of WWW Latency, INET 97, 1997, http://wwwiscoorg/inet97/procceding/a1/a1_3htm [4] Arthur Goldberg, Ilya Pevzner and Robert Buff, Caching Characteristic of Internet and Intranet Web proxy Traces, In Computer Measurement Group Conference (CMG 98), Anaheim, CA, December 1998, http://wwwcsnyuedu/artg [5] David Barnes and Neil G Smith, An Analysis of World-Wide Web Proxy Cache Performance and its Application to the Modelling and Simulation of Network Traffic, In Proceedings of the Fourth International Conference on Telecommunication Systems Modeling and Analysis, March 1996, http://wwwcsukcacuk/people/staff/djb/pubshtml [6] Themistoklis Palpanas and Alberto Mendelzon, Web Prefetching Using Partial Match Prediction, In Web Caching Workshop WCW 99, 1999, http://wwwircachenet/cache/workshop99/programhtml [7] Squid Internet Object Cache, available from http://squidnlanrnet/squid/ [8] Gihan VDias, Graham Cope and Ravi Wijayaratne, A Smart Internet Caching System, INET 96 Conference, 1996, http://wwwisocorg/isoc/ whatis/conferences/inet/96/proceedings/a4/a4_3htm [9] Katsuo Doi, WWW Access by Proactively Controlled Caching Proxy, Sharp Technical Journal, No 66, December 1996 [10] Brad Duska, David Marwood, and Michael JFeeley, The Measured Access Characteristics of World-Wide Web Client Proxy Caches, In Usenix Symposium on Internet Technologies and Systems (USITS), Monterey, CA, USA, December 8-11 1997, Usenix, http://wwwcsubcca/spider/marwood/ Projects/SPA/Report/Reporthtml 57

[11] Marc Abrams, CRStandridge, GAbdulla, SWilliams, and EAFox, Caching Proxies: Limitations and Potentials, In Proceedings of the Fourth International WWW Conference, 1995, http://eicsvtedu/~succeed/www4/ WWW4html [12] Anawat Chankhunthod et al, A Hierarchical Internet Object Cache, Technical Conference, Usenix 1996, http://excaliburuscedu/cachehtml/cachehtml [13] James Griffioen and Randy Appleton Reducing File System Latency using a Predictive Approach, Proceddings of the 1994 Summer USENIX Technical Conference, Boston, Massachusetts, USA, 1994, http://usenixorg/publications/library/proceedings/bos94/griffioenhtml [14] Azer Bestavros, Speculative Data Dissemination and Service to Reduce Server Load, Network Traffic and Service Tome in Distributed Information System, In International Conference on Data Engineering, pages 180-189, New Orleans, LO, February 1996 [15] Tomas M Kroeger, Darrell D E Long, and Jeffrey C Mogul Exploring the bounds of web latency reduction from caching and prefetching, In Proceedings of USENIX Symposium on Internet Technology and Systems, December 1997, http://wwwusenixorg/publications/library/proceedings/ usits97/kroegerhtml [16] Evangelos P Margatos and Catherine E Chronaki, A top-10 Approach to Prefetching on the Web, Technical report, In Proceedings of INET' 98 (The Internet Summit), Geneva, Switzerland, July 1998, http://wwwicsforthgr/ proj/arch-vlsi/os/wwwhtml [17] Wcol Group, WWW Collector the prefetching proxy server for WWW, 1997, http://shikaaist-naraacjp/products/wcol/wcolhtml [18] Li Fan, Quinn Jacobson, Pei Cao, and Wei Lin, Web Prefetching Between Low-Bandwidth Clients and Proxies: Potential and Performance, In Proceedings of the Joint International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS '99), Atlanta, GA, May 1999, http://wwwcswiscedu/~cao/ [19] Wget, available from http://subzerocampusluthse/freedocs/wget- 142/wget_tochtml [20] Tim Bray, Measuring the Web, In Proceedings of the Fifth International 58

World Wide Web Conference, pages 993-1005, Paris, France, May 1996 [21] Allison Woodruff, Paul M Aoki, Eric Brewer, Paul Gauthier, and Lawrence A Rowe, An Investigation of Documents from the WWW, In Proceedings of the Fifth International WWW Conference, pages 963-979, Paris, France, May 1996 [22] T Berners-Lee, R Fielding, and H Frystyk, Hypertext Transfer Protocol HTTP/10, RFC 1945, May, 1996 [23] R Fielding, J Gettys, J Mogul, H Frystyk, L Masinter, P Leach and T Berners-Lee, Hypertext Transfer Protocol - HTTP/11, RFC 2616, June 1999 [24] Evangelos Markatos, Catherine E Chronaki, A Top-10 Approach to Prefetching on the Web, Technical Report No 173, ICS-FORTH, Heraklion, Crete, Greece,August 1996 [25] Ghaleb Abdulla, Edward A Fox, Marc Abrams, and Stephen Williams, WWW Proxy Traffic Characterization with Application to Caching, Technical Report TR-97-03, Computer Science Department, Virginia Tech, March 1997, http://wwwcsvtedu/~chitra/workhtml [26] James E Pitkow, Summary of WWW characterizations, In Proceedings of the Seventh International World Wide Web Conference, Brisbane, Australia, April 1998, http://www7scueduau/programme/fullpapers/1877/ com1877htm [27] Pei Cao, Edward W Felten, Anna R Karlin, and Kai Li, A Study of Integrated Prefetching and Caching Strategies, In Proceedings of the ACM SIGMETRICS Conference on Measurement and Modeling of Computer Systems, May 1995, http://wwwcswiscedu/~cao/publicationshtml [28] Eric A Brewer, Paul Gauthier, and Dennis McEvoy, The long-term viability of large-scale caching, In Proceedings of the Third International WWW Caching Workshop, Manchester, England, June 1998, http://wwwcachejanet/events/workshop [29] CacheRaQ of Cobalt Network, available from http://wwwcoblatnetcom [30] Webspeed of Packetstorm Technologies, available from http://wwwpacketstormonca/products/webspeed/featuresindetailhtml [31] Bert Williams, Transparent web caching solutions, In Proceedings of the Third International WWW Caching Workshop, Manchester, England, June 59

1998 http://wwwcachejanet/events/workshop/33/cachepaperhtml [32] David H Crocker, Standard For The Format Of Arpa Internet Text Message, RFC 822, August 13, 1982 [33], Protocals for the World-Wide Web, available form http://pecetrirekr/~qkim/http/c11html 60

61