SAP HANA vs Oracle - Speaker Deck

Slide 1

Slide 1 text

੿࢚ળ HANA vs.

Slide 2

Slide 2 text

ࣽࢲ

Slide 3

Slide 3 text

ࣽࢲ HANA

Slide 4

Slide 4 text

ࣽࢲ HANA = In Memory DB

Slide 5

Slide 5 text

ࣽࢲ HANA Databaseо ו۰ࢲ = In Memory DB

Slide 6

Slide 6 text

ࣽࢲ HANA Database = = In Memory DB

Slide 7

Slide 7 text

ࣽࢲ HANA Database HDD ࠁ׮ ࡅܲ RAM = = In Memory DB

Slide 8

Slide 8 text

ࣽࢲ HANA Database HDD ࠁ׮ ࡅܲ RAM ੷੢੢஖੄ ৉ࢎ = = = In Memory DB

Slide 9

Slide 9 text

= = = ࣽࢲ HANA HDD ࠁ׮ ࡅܲ RAM Database ੷੢੢஖੄ ৉ࢎ ࢸݺ ࣽࢲ In Memory DB

Slide 10

Slide 10 text

1. DBח ৵ కয լաਃ?

Slide 11

Slide 11 text

ীפঈ੄ ١੢ = ҅࢑ӝ Reference: http://ftp.arl.mil/ftp/historic-computers/, Copyright Rovio Entertainment Ltd.

Slide 12

Slide 12 text

ীפঈ੄ ١੢ = ҅࢑ӝ Reference: http://ftp.arl.mil/ftp/historic-computers/, Copyright Rovio Entertainment Ltd.

Slide 13

Slide 13 text

No content

Slide 14

Slide 14 text

Data

Slide 15

Slide 15 text

Data Data Data

Slide 16

Slide 16 text

ؘ੉ఠ ૐо = DB ࣻਃ Copyright (C) Apple Inc., Template - CoreData.icns

Slide 17

Slide 17 text

Ӓؘ۠ࢿמ੉݅઒झۣ૑ঋই৵Ӓ۠Ѧө

Slide 18

Slide 18 text

2. ੷੢੢஖੄ ৉ࢎ

Slide 19

Slide 19 text

زҷ ߷ച Reference: http://en.wikipedia.org/wiki/File:AltamiraBison.jpg

Slide 20

Slide 20 text

੗ӝ ప੉೐ Reference: http://en.wikipedia.org/wiki/File:Quarter-Inch_Cartridges.jpg

Slide 21

Slide 21 text

ೞ٘ ٣झ௼ Reference: http://en.wikipedia.org/wiki/File:Seagate_ST33232A_hard_disk_inner_view.jpg

Slide 22

Slide 22 text

۔ Reference: http://en.wikipedia.org/wiki/File:4GB_DDR3_SO-DIMM.jpg

Slide 23

Slide 23 text

ޙઁ = וܽ HDD

Slide 24

Slide 24 text

HDDо ঴݃ա וܽоਃ

Slide 25

Slide 25 text

HDDо ঴݃ա וܽоਃ ؀৉಩ ߈਽ ࣘب CPU GPU RAM RAM HDD 25.6GB/s 0.32ns 153GB/s 0.83ns 17.06GB/s 6ns 120MB/s 4.16ms

Slide 26

Slide 26 text

HDDо ঴݃ա וܽоਃ 0 50 100 150 200 CPU GPU RAM HDD 0 1.5 3.0 4.5 6.0 ؀৉಩ ߈਽ ࣘب GB/s ns

Slide 27

Slide 27 text

׹׹ೞפө

Slide 28

Slide 28 text

؀৉಩ = ੍ӝ ࣘب

Slide 29

Slide 29 text

Contributed by AvatarP

Slide 30

Slide 30 text

؀৉಩ = ੍ӝ ࣘب 3.63GB LEGENDS LEAGUEof League of Legends and Riot Games Inc. are trademarks or registered trademarks of Riot Games, Inc.

Slide 31

Slide 31 text

؀৉಩ = ੍ӝ ࣘب 30ୡ 3.63GB LEGENDS LEAGUEof League of Legends and Riot Games Inc. are trademarks or registered trademarks of Riot Games, Inc.

Slide 32

Slide 32 text

؀৉಩ = ੍ӝ ࣘب 30ୡ 0.21ୡ 3.63GB LEGENDS LEAGUEof League of Legends and Riot Games Inc. are trademarks or registered trademarks of Riot Games, Inc.

Slide 33

Slide 33 text

؀৉಩ = ੍ӝ ࣘب 30ୡ 0.21ୡ 3.63GB LEGENDS LEAGUEof League of Legends and Riot Games Inc. are trademarks or registered trademarks of Riot Games, Inc. ߓ

Slide 34

Slide 34 text

߈਽ ࣘب CPU HDD

Slide 35

Slide 35 text

߈਽ ࣘب CPU ੗ܐખ ׮য় HDD

Slide 36

Slide 36 text

߈਽ ࣘب CPU ੗ܐખ ׮য় ܻ٘ѷणפ׮ HDD

Slide 37

Slide 37 text

߈਽ ࣘب CPU ੗ܐખ ׮য় ܻ٘ѷणפ׮ HDD 4.16ms

Slide 38

Slide 38 text

߈਽ ࣘب CPU ੗ܐખ ׮য় ܻ٘ѷणפ׮ RAM

Slide 39

Slide 39 text

߈਽ ࣘب CPU ੗ܐખ ׮য় ܻ٘ѷणפ׮ RAM 6ns

Slide 40

Slide 40 text

HDD RAM ߈਽ ࣘب

Slide 41

Slide 41 text

HDD RAM 4.16ms 6ns ߈਽ ࣘب

Slide 42

Slide 42 text

HDD RAM 4.16ms 6ns 1ୡ ߈਽ ࣘب

Slide 43

Slide 43 text

HDD RAM 4.16ms 6ns 1ୡ 8ੌ ߈਽ ࣘب

Slide 44

Slide 44 text

HDD RAM 4.16ms 6ns 1ୡ 8ੌ ߈਽ ࣘب ߓ

Slide 45

Slide 45 text

ஹೊఠ ҳઑ

Slide 46

Slide 46 text

ஹೊఠ ҳઑ CPU

Slide 47

Slide 47 text

ஹೊఠ ҳઑ CPU GPU

Slide 48

Slide 48 text

ஹೊఠ ҳઑ CPU GPU Chipset

Slide 49

Slide 49 text

ஹೊఠ ҳઑ CPU GPU 16GB/s (PCIe v3.0) Chipset

Slide 50

Slide 50 text

ஹೊఠ ҳઑ CPU GPU 16GB/s (PCIe v3.0) Chipset 17GB/s (DDR3-2133)

Slide 51

Slide 51 text

ஹೊఠ ҳઑ CPU GPU 16GB/s (PCIe v3.0) 25.6GB/s ([email protected]) Chipset 17GB/s (DDR3-2133)

Slide 52

Slide 52 text

ஹೊఠ ҳઑ CPU GPU 16GB/s (PCIe v3.0) 25.6GB/s ([email protected]) Chipset 17GB/s (DDR3-2133) 0.1GB/s

Slide 53

Slide 53 text

ஹೊఠ ҳઑ CPU GPU 16GB/s (PCIe v3.0) 25.6GB/s ([email protected]) Chipset 17GB/s (DDR3-2133) 0.1GB/s ߽ݾഅ࢚

Slide 54

Slide 54 text

)%%ܳॳח੉ਬח

Slide 55

Slide 55 text

࠺ऴ3". )%%ܳॳח੉ਬח

Slide 56

Slide 56 text

3. ڄয૑ח ݫݽܻ оѺ

Slide 57

Slide 57 text

1955 109 1 10-5 $/1MB Reference: http://www.jcmit.com/mem2012.htm 2010

Slide 58

Slide 58 text

1955 109 1 10-5 $/1MB Reference: http://www.jcmit.com/mem2012.htm 2010 DIMM SIMM IC

Slide 59

Slide 59 text

1955 109 1 10-5 $/1MB Reference: http://www.jcmit.com/mem2012.htm 2010 DIMM SIMM IC Flash

Slide 60

Slide 60 text

1955 109 1 10-5 $/1MB Reference: http://www.jcmit.com/mem2012.htm 2010 DIMM SIMM IC Flash HDD

Slide 61

Slide 61 text

1955 109 1 10-5 $/1MB Reference: http://www.jcmit.com/mem2012.htm 2010 DIMM SIMM IC Flash HDD ݆੉ ղ۷׮

Slide 62

Slide 62 text

Ӓېࢲ աৡѪ੉

Slide 63

Slide 63 text

In Memory DB

Slide 64

Slide 64 text

In Memory DB

Slide 65

Slide 65 text

In Memory DB Data Data

Slide 66

Slide 66 text

In Memory DB Data Data

Slide 67

Slide 67 text

In Memory DB • ݫݽܻо ऱ૑פө • ݫݽܻо ࡅܰפө • ݆਷ ؘ੉ఠ ૊द ࠙ࢳ

Slide 68

Slide 68 text

৘ܳ ٜযࠁݶ • DB • ݽٚ द೷ ੿ࠁ • ١ࣻо ҾӘ

Slide 69

Slide 69 text

৘ܳ ٜযࠁݶ • ա੄ ࢿ੸ ઑഥ • ࣻъ ղ৉ оઉয়ӝ • ࣻъ ղ৉ ߹ ੼ࣻ • ١ࣻ ҅࢑ೞӝ • ࠁৈ઱ӝ Data Data

Slide 70

Slide 70 text

৘ܳ ٜযࠁݶ • ա੄ ࢿ੸ ઑഥ • ݒ द೷ 10ޙઁ * 2ߣ • ೠ ೟ӝ 6җݾ • 8೟ӝ • ࣻъࢤ 20ݺ • 10*2*6*8*20*4.16ms = 79ୡ Data Data

Slide 71

Slide 71 text

예를 들어 일본 노무라연구소는 인메모리 컴퓨팅을 이용해 도쿄에서 운행하고 있는 1만2000대 택시에서 발생하는 교통정보를 1초에 분석해 최단 이동경로를 운전자들에게 제공한다. 교통흐름을 실시간으로 분석 해 막히지 않는 구간을 안내하는 것. 택시들이 최단 이동 경로로 움직이기 때문에 교통 혼잡으로 인한 손실을 막을 수 있다. 또 단순히 택시 회사의 이익뿐 아니라 교통 분산 효과를 가져오기 때문에 공공인프라 개선이라는 부수적 효과도 가져온다. Reference: http://www.ddaily.co.kr/news/news_view.php?uid=99448

Slide 72

Slide 72 text

Slide 73

Slide 73 text

실시간 처리 실시간 분석 실시간 예측 인공지능 사용자 프로파일링 무서운 미래

Slide 74

Slide 74 text

ORLANDO, Fla. — HP and SAP AG (NYSE:SAP) today announced advancements to a joint initiative, codenamed “Project Kraken,” to improve business processes and mitigate risks for customers. ... Project Kraken’s goal is to provide customers additional choice in scalability for large, online transaction processing applications, with requirements for real-time analytical insights into this data. This allows organizations, including retailers, financial institutions, utilities and governments, another alternative to quickly detect patterns, analyze massive data volumes on the fly, and perform their operations quickly. The prototype system, optimized with 16 Intel® Xeon® processor E7 family (codenamed Ivy Bridge-EX) central processing units (CPUs) and 12 terabytes of memory, has been engineered specifically for complex, data-intensive workloads. These include supply chain, customer relationship management (CRM), enterprise resource planning (ERP) and data analysis. HP and SAP chose Intel processors because of the company’s leading-edge technology that enables customers to improve system performance. HP and SAP Advance SAP HANA through Joint Innovation SAP, SAP HANA and all SAP logos are trademarks or registered trademarks of SAP AG in Germany and in several other countries. Intel and Xeon are trademarks of Intel Corporation in the U.S. and other countries.

Slide 75

Slide 75 text

• Project Kraken • ؀ഋ DBীࢲ पदр ࠙ࢳ • ಁఢ ଺ӝ, ؀ਊ۝ ؘ੉ఠ पदр ࠙ࢳ ١ • 16 Intel® Xeon® processor E7 • 12 TB ݫݽܻ (12,228 GB) HP ৬ SAP ഈ۱

Slide 76

Slide 76 text

Apple NAB Bay (2007), Photo from http://sysadminporn.com/wp-content/uploads/2011/01/Xserve-HD_2560.jpg

Slide 77

Slide 77 text

1 U =

Slide 78

Slide 78 text

1 U = 44.45mm

Slide 79

Slide 79 text

No content

Slide 80

Slide 80 text

Apple XServe, http://www.creativeplanetnetwork.com/videography/uploadedimages/VideoGraphy/Videography_News/0408AppleXserve-1bg.jpg 79cm 48cm

Slide 81

Slide 81 text

SuperComputing 2012, IBM Power7, Photo from http://hpc.uark.edu/hpc/blog/sc12.html

Slide 82

Slide 82 text

4. Ӓٜ਷ ־ҳੋо

Slide 83

Slide 83 text

Public Domain by amadeusm

Slide 84

Slide 84 text

• 1972֙ ةੌীࢲ ࢸ݀ • ERPҙ۲ (੹ࢎ੸ ੗ਗ ҙܻ) • ERPীח DBо ೙ਃ

Slide 85

Slide 85 text

द੢ ੼ਬਯ SAP Salesforce Oracle Microsoft Amdocs Other Source: Market Share Analysis: ERP Software, Worldwide, 2012, http://www.forbes.com/sites/louiscolumbus/2012/12/04/roundup-of-crm-forecasts-and-market-estimates-2012/

Slide 86

Slide 86 text

द੢ ੼ਬਯ 19% SAP Salesforce Oracle Microsoft Amdocs Other Source: Market Share Analysis: ERP Software, Worldwide, 2012, http://www.forbes.com/sites/louiscolumbus/2012/12/04/roundup-of-crm-forecasts-and-market-estimates-2012/

Slide 87

Slide 87 text

Video: http://youtu.be/KT5XMBzWpMc

Slide 88

Slide 88 text

Reference: http://www.ﬂickr.com/photos/93755244@N00/4140459965 by dahlstroms

Slide 89

Slide 89 text

• 1977֙ ޷Ҵীࢲ ࢸ݀ • ഥࢎਊ H/W, S/W (ౠ൤ DB) • ݆਷ ഥࢎ ੋࣻ (85 + 11 ੉࢚) • SUN(Java) - $7.4B @ 2010 #3 • MySQL - ($1.0B @ 2008 #9) • PeopleSoft - $10.3B @ 2005 #1

Slide 90

Slide 90 text

Oracle DB IBM DB2 MS SQL SAP/Sybase Teradata Others द੢ ੼ਬਯ Source: Gartner 2012, RDBMS Software Revenue, http://itknowledgeexchange.techtarget.com/eye-on-oracle/oracle-the-clear-leader-in-24-billion-rdbms-market/

Slide 91

Slide 91 text

49% 20% 17% 5% 4% 6% Oracle DB IBM DB2 MS SQL SAP/Sybase Teradata Others द੢ ੼ਬਯ Source: Gartner 2012, RDBMS Software Revenue, http://itknowledgeexchange.techtarget.com/eye-on-oracle/oracle-the-clear-leader-in-24-billion-rdbms-market/

Slide 92

Slide 92 text

$12B $5B $4B $1B $1B $1B Oracle DB IBM DB2 MS SQL SAP/Sybase Teradata Others DB ݒ୹ Source: Gartner 2012, RDBMS Software Revenue, http://itknowledgeexchange.techtarget.com/eye-on-oracle/oracle-the-clear-leader-in-24-billion-rdbms-market/

Slide 93

Slide 93 text

No content

Slide 94

Slide 94 text

 ஝ OS  339S0073ARM K2132C2PD-50-F ONO650908 APL0298 N1PVNMPP 0919 iPhone OS

Slide 95

Slide 95 text

 ஝ OS TIZEN ANDROID PA SAMI

Slide 96

Slide 96 text

 ஝ OS OS ஝ ҃੬ ҙ҅

Slide 97

Slide 97 text

DB ERP PeopleSoft Transaction In Memory -Oracle이 PeopleSoft를 2005년에 $10,300M (103억 달러 = 11조...) -SAP의 TIM인수는 2005년. 가격은 안밝 혀짐

Slide 98

Slide 98 text

DB ERP -Oracle이 PeopleSoft를 2005년에 $10,300M (103억 달러 = 11조...) -SAP의 TIM인수는 2005년. 가격은 안밝 혀짐 DB ERP ҃੬ ҙ҅

Slide 99

Slide 99 text



Slide 100

Slide 100 text

Slide 101

Slide 101 text

Ѿۿ

Slide 102

Slide 102 text

Ѿۿ • In Memory DBח ޖ૑ ࡅܰ׮.

Slide 103

Slide 103 text

Ѿۿ • In Memory DBח ޖ૑ ࡅܰ׮. • पदрਵ۽ ୊ܻೡ ࣻ ੓׮.

Slide 104

Slide 104 text

Ѿۿ • In Memory DBח ޖ૑ ࡅܰ׮. • पदрਵ۽ ୊ܻೡ ࣻ ੓׮. • ݫݽܻח ষ୒ ऱ઎׮.

Slide 105

Slide 105 text

Ѿۿ • In Memory DBח ޖ૑ ࡅܰ׮. • पदрਵ۽ ୊ܻೡ ࣻ ੓׮. • ݫݽܻח ষ୒ ऱ઎׮. • Oracleب ௼Ҋ SAPب ௼׮. ل੉ ҃੬ೠ׮.

Slide 106

Slide 106 text

Ѿۿ • In Memory DBח ޖ૑ ࡅܰ׮. • पदрਵ۽ ୊ܻೡ ࣻ ੓׮. • ݫݽܻח ষ୒ ऱ઎׮. • Oracleب ௼Ҋ SAPب ௼׮. ل੉ ҃੬ೠ׮. • ৈ۞࠙ب જ਷Ѫ ٜ݅য ੜ ౹ইࠁ੗.

Slide 107

Slide 107 text

Q&A