Hannover Messe 2024: Generate business value from your SAP data

AWS Events
28 May 202425:05

Summary

TLDRこのビデオスクリプトでは、製造業向けのSnowflake for SAPデータ分析についてDavid Rickerが語っています。市場動向としてM&Aの増加とそれに伴う迅速な財務報告の必要性、BW(Business Warehouse)の廃止とその遅れ、顧客が求めるイノベーションとスケーラビリティ、データコラボレーションの重要性が説明されています。SnowflakeはSAPシステムと異なるデータソースを統合し、大規模なデータを効率的に分析・管理できるプラットフォームとして紹介されています。また、Snowflakeのパートナーシップと生態系、特にAWSとの連携が強調されています。スクリプトは、製造業におけるデータ分析の課題とSnowflakeが提供する解決策、そして実際の顧客事例を通じてその価値を説くものとなっています。

Takeaways

  • 📈 スノウフレークは製造業向けのSAPデータ分析に適しており、企業は急速に統合や分割を遂げ、財務諸表を迅速に作成する必要があると述べています。
  • 🔄 市場動向として、企業はM&Aを通じて急速に拡大し、SAPシステムが異なるブループリントを持つ多数の企業が統合されるケースが増えています。
  • 📊 SAPのBWシステムは徐々に廃止されており、2029年までに段階的に切られる予定です。顧客はイノベーション、スケーラビリティ、安定性、データコラボレーションを求めています。
  • 🚀 スノウフレークは最大40ペタバイトのデータを扱える大規模なデータベースであり、SAPの最大BWシステムの277倍以上のスケールを提供しています。
  • 🔗 スノウフレークはデータコラボレーションを促進し、サプライチェーンにおける顧客とのデータ共有を容易にします。
  • 💡 ERPとSAPデータを統合し、分析するための課題は、多数のERPシステムからデータを集約し、効率的に分析する必要性です。
  • 🛠️ スノウフレークは非構造化データや半構造化データを容易に取り込み、管理し、分析することができます。
  • 💼 スノウフレークはビジネス価値の迅速提供に貢献し、エンタープライズの可視性を高めるだけでなく、ビジネス改善プロセスを強化します。
  • 🌐 スノウフレークはクラウドベースのシステムであり、パートナーシップのエコシステムを通じて顧客間のデータ共有を可能にしています。
  • 🔧 スノウフレークはAWS、Microsoft、Googleなどとの互換性があり、様々な技術パートナーシップを通じて強化されています。
  • 🔄 大規模企業はSAP BWからスノウフレークへの移行を通じて、コスト削減と効率化を実現しています。

Q & A

  • スノーフレークは製造業向けのSAPデータ分析にどのように役立つか説明してください。

    -スノーフレークは製造業向けにSAPデータ分析を提供し、企業が迅速な合併や事業譲渡に対応する際に財務諸表を迅速に作成し、データの統合と分析を容易に行うことを支援します。

  • スノーフレークが提供する市場動向にはどのようなものがありますか?

    -スノーフレークは市場からの要望に応えるために、企業がより迅速に合併や事業譲渡を行う必要があるという動向に加え、SAP BWのフェーズアウトとそれに伴う顧客のイノベーションニーズに対応しています。

  • スノーフレークはSAP BWと比較してどのような利点がありますか?

    -スノーフレークはSAP BWよりもスケーラブルで安定しており、大量のデータを扱うことができる利点があります。スノーフレークは最大40ペタバイトのデータを扱うことができ、SAP BWの最大容量をはるかに上回ります。

  • スノーフレークにおけるデータコラボレーションとは何を意味しますか?

    -データコラボレーションは、企業がサプライチェーンにおける顧客やパートナーとデータを共有することができることを意味しており、ネットワークを介してデータを移動する必要なく、ペタバイト規模のデータを効率的に共有できます。

  • スノーフレークはどのような課題を解決するのに役立ちますか?

    -スノーフレークはERPシステムからのデータの統合、分析、およびAI/MLの活用を支援し、企業が経済的な方法でデータをエンリッチし、効率的に分析することができるようにします。

  • スノーフレークにおけるゼロコピークローンとは何ですか?

    -ゼロコピークローンはスノーフレークの機能で、データをコピーする際に追加のストレージを必要とせず、同じサイズのデータを別の環境で使用可能にします。これにより、開発やテスト環境を効率的に作成・破棄できます。

  • スノーフレークはどのようなビジネスプロセス改善に貢献していますか?

    -スノーフレークはエンタープライズの可視性を高めるだけでなく、ビジネス改善プロセスに貢献しています。それは予測能力の強化、供給連のリスク管理、在庫可用性の改善などを通じて行われます。

  • スノーフレークはどのようなパートナーと提携していますか?

    -スノーフレークはAWS、Microsoft、Googleなどのクラウドプロバイダーと提携しており、さらにシステムインテグレーターやデータサービスプロバイダーと協力して顧客のニーズを満たしています。

  • スノーフレークにおけるSAPデータアーキテクチャの代替としてどのような機能がありますか?

    -スノーフレークはSAPデータアーキテクチャの代替として、データを取り込み、モデル化し、レポート作成することができる機能を提供しています。これは従来のSAP BWやHANAアーキテクチャとは異なるアプローチです。

  • スノーフレークはどのような技術的なユースケースを解決していますか?

    -スノーフレークはSAP Enterprise Data Warehouseの代替としてコストを大幅に削減し、またSAP BWからワークロードを移動してコストを節約するのに役立ちます。さらに、リアルタイムでのデータの取り扱いも可能です。

  • スノーフレークはどのようなSAPデータを取り扱うことができますか?

    -スノーフレークはSAPから提供される様々なデータを取り扱うことができます。それは透明テーブル、プールテーブル、クラスターテーブルなどを含むSAPのデータを指し示します。

  • スノーフレークでのリアルタイムデータ分析はどのように実現されますか?

    -スノーフレークではトリガーベース、バッチベース、マイクロバッチベースの方法でリアルタイムデータ分析を実現しており、さまざまなパートナー企業がこのプロセスをサポートしています。

  • スノーフレークにおけるデータのセキュリティはどのように確保されていますか?

    -スノーフレークはデータのセキュリティを確保するために包括的なセキュリティ機能を提供しており、これはデータの取り扱い、格納、およびアクセスを管理する機能を含みます。

  • スノーフレークはどのようなSAPサポートツールを使用していますか?

    -スノーフレークはSAPサポートツールとしてデータサービスを使用しており、これらのツールはSAPとスノーフレークの間のデータを効率的に移動させるために設計されています。

  • スノーフレークでのデータ分析にはどのような方法がありますか?

    -スノーフレークでは、SQLだけでなくPython、Scala、Javaなどのプログラミング言語を使用してデータ分析を行えます。また、機械学習やAIのライブラリも利用可能です。

  • スノーフレークでのデータ分析の利点は何ですか?

    -スノーフレークでのデータ分析の利点は、モデルに依存しない柔軟性、豊富な分析機能、および他のツールとの統合性です。これにより、複雑なデータ分析を効率的に行うことができます。

  • スノーフレークはどのような顧客事例を挙げていますか?

    -スノーフレークは顧客事例として、SAPデータをスノーフレークに移行し、分析ユーザー数を数千人規模に拡大する企業を挙げています。これにより、供給リスクの低減やコスト効率の向上が実現されています。

  • スノーフレークでの分析環境の自動化とはどのようなものですか?

    -スノーフレークでの分析環境の自動化とは、ユーザー向けのスノーフレーク環境の自動生成と展開を意味しており、これは従来の手動でのプロセスよりも効率的でスケーラブルです。

  • スノーフレークへの移行の最初のステップは何ですか?

    -スノーフレークへの移行の最初のステップは、現状の分析を評価し、使用されていない機能を特定することであり、その後必要な機能のみをスノーフレークに移行する戦略を立てます。

Outlines

00:00

📊 SAP データ分析に関する市場動向

スノーフレークのDavid Rickerは、製造業向けのSAPデータ分析について語ります。市場の力学やCOVID-19後の企業の合併・分割がデータに与える影響について説明し、SAPシステムの統合と財務データの統合の難しさを強調します。また、BWシステムの段階的廃止と、それに対する顧客のニーズについても触れています。さらに、SAPのデータ量の増加とそれに対応するための技術的な課題についても議論しています。

05:02

🔍 データ統合と管理の重要性

Davidは、SAPシステムとスノーフレークの統合について詳述します。データの統合、ガバナンス、中央集権化の重要性を強調し、企業がより効果的にデータを活用するための方法を紹介します。AIやMLを活用したデータの強化や、SAPデータと非構造化データの統合の必要性についても説明しています。

10:04

🌐 ビジネスプロセスの改善と予測強化

ビジネスプロセスの改善と予測の強化について話します。さまざまなデータソースを統合することで、税務情報やサプライチェーンリスクなどの重要なビジネス質問に答える能力を高めます。システムの高可用性を維持しながら、効率的なデータ処理を実現するためのスノーフレークの利点を強調しています。

15:06

🤝 パートナーシップとデータ共有の利点

スノーフレークのエコシステムとパートナーシップの重要性について議論します。クラウドベースのシステムにより、データの共有が容易になり、サプライチェーン全体での効率が向上します。さらに、SAPシステムとの統合と、データのリアルタイム取得に関する技術的な側面についても説明しています。

20:09

🚀 AWSとの統合による効率向上

製造業向けのAWSとの統合アーキテクチャの利点について解説します。経済的で効率的なデータ統合を実現するための方法を示し、スノーフレークのスケーラビリティとコスト効率の良さを強調しています。また、複雑なSAPシステムの統合を簡素化するための具体的なステップや、ビジネス上の利点についても触れています。

Mindmap

Keywords

💡Snowflake

Snowflakeは、クラウドベースのデータウェアハウスソリューションです。ビデオでは、製造業界におけるSAPデータ分析にSnowflakeを使用する利点を強調しており、そのスケーラビリティと経済性について説明しています。例えば、スクリプトではSnowflakeが最大40エクサバイトのデータを扱うことができると述べており、これは従来のSAP BWシステムと比較して277倍大きい容量を持つと比較しています。

💡SAP BW

SAP BWは、SAP社が提供するデータウェアハウスソフトウェアです。ビデオでは、SAP BWが徐々に廃止されつつあることと、その代替としてSnowflakeが提供する利点を説明しています。スクリプトでは、SAP BWの廃止予定が2029年と延びており、顧客が革新、サイズの可変性、安定性、データコラボレーションを求めていると指摘しています。

💡データ分析

データ分析とは、大量のデータを収集、クリーニング、解釈し、ビジネス上の洞察を得るプロセスを指します。ビデオでは、Snowflakeを利用して製造業界の企業がより迅速かつコスト効率の高いデータ分析を行うことができると述べています。スクリプトからの例として、企業がSnowflakeを利用して生産停止や遅延と関連する顧客にリアルタイムで連絡を取る方法が紹介されています。

💡M&A

M&Aは、企業の合併と買収を意味します。ビデオでは、企業がM&Aを通じて急速に拡大し、それに伴い財務諸表を迅速に統合する必要があると説明しています。スクリプトでは、企業がM&Aを繰り返し行い、30または40社のSAPシステムを統合する例が挙げられており、そのデータ統合の課題がSnowflakeによって解決されると述べています。

💡データコラボレーション

データコラボレーションは、異なる組織間でデータを共有し、ビジネスの価値を創出するプロセスです。ビデオでは、Snowflakeがデータをペタバイト単位で共有できる機能を持っていると強調しており、これはサプライチェーンにおけるデータの迅速な共有に不可欠です。スクリプトでは、データコラボレーションがSAPの顧客が求める重要な要素の1つとなっていると述べています。

💡スケーラビリティ

スケーラビリティは、システムやアプリケーションが成長するニーズに応じて拡張できる能力を指します。ビデオでは、Snowflakeが非常にスケーラブルであると述べており、これはSAP BWシステムと比較して277倍のデータ容量を持つことができることを意味します。スクリプトでは、Snowflakeのスケーラビリティが製造業界の企業にとって重要な利点であると強調しています。

💡データウェアハウス

データウェアハウスは、ビジネスインテリジェンスのために集約されたデータの格納先です。ビデオでは、Snowflakeがデータウェアハウスとして機能し、さまざまなデータソースからのデータを統合し、分析するためのプラットフォームを提供していると説明しています。スクリプトでは、Snowflakeがデータウェアハウスとして機能し、SAP BWの代替として優れた選択肢であると述べています。

💡SAP S/4 HANA

SAP S/4 HANAは、SAP社の次世代ERPソリューションです。ビデオでは、SAP S/4 HANAが新しいビジネスアプリケーションのためのプラットフォームとして位置づけられており、SAP BWから計画機能をSAP Analytics Cloudへ、集約機能をS/4 HANAへ移行していると説明しています。スクリプトでは、SAP S/4 HANAがSnowflakeと組み合わせることで、より効率的なデータ管理が可能になると述べています。

💡マージ

マージは、複数のデータセットを統合するプロセスです。ビデオでは、企業が複数のSAPシステムをマージし、統一された財務諸表を作成する必要があると説明しています。スクリプトでは、Snowflakeがこのようなマージプロセスを効率化し、データの迅速な統合と分析が可能になると述べています。

💡分散型アーキテクチャ

分散型アーキテクチャは、データや処理を複数の場所に分散させる設計です。ビデオでは、Snowflakeが分散型アーキテクチャを採用しており、これはデータの冗長性と信頼性を高めるだけでなく、スケーラビリティも確保する利点があると説明しています。スクリプトでは、Snowflakeの分散型アーキテクチャがシステムの高可用性と柔軟なスケーリングを可能にしていると述べています。

Highlights

Snowflakeは製造業向けのSAPデータ分析に大きな進展をもたらします。

Snowflakeは最大40ペタバイトのデータを扱うことができ、SAP BWの最大システムを大幅に上回ります。

Snowflakeは、異なるSAPシステム間でのデータ統合を容易にします。

SAP BWは長年にわたって段階的に廃止されていますが、Snowflakeは新しいデータ管理ソリューションを提供します。

Snowflakeは、企業が迅速に財務データを統合し、分析するための強力なツールを提供します。

データのコラボレーションと共有が容易で、特にサプライチェーンの効率化に役立ちます。

Snowflakeのゼロコピークローン機能により、開発とテストのためのシステムを迅速に作成できます。

Snowflakeは、従来のETLプロセスを簡素化し、データレイクとデータウェアハウスの両方の機能を兼ね備えています。

大規模な企業は、Snowflakeを使用して数百のERPシステムを統合し、分析ユーザーの数を大幅に増加させています。

Snowflakeは、AIおよびML機能を活用して、データの価値を高めます。

Snowflakeの高可用性と冗長性により、システムのダウンタイムを最小限に抑えることができます。

Snowflakeは、コスト効率の高いデータストレージおよび分析ソリューションを提供します。

Snowflakeのパートナーエコシステムにより、さまざまなデータソースからの統合が容易になります。

Snowflakeは、製造業におけるデータ統合と分析を大幅に簡素化します。

Snowflakeは、SAPシステムとその他のデータソースからのデータを効率的に統合します。

Snowflakeは、企業が迅速に市場変化に対応できるよう支援します。

Transcripts

play00:00

[Music]

play00:02

[Applause]

play00:02

[Music]

play00:10

[Applause]

play00:11

the next couple minutes I'll be talking

play00:13

about the snowflake for manufacturing

play00:15

sap data analytics I'm David Ricker uh

play00:19

part of snowflake part of the field CTO

play00:22

office based out of

play00:26

Geneva uh we have some future statements

play00:29

so have a safe harbor

play00:32

piece so if we take a look at what some

play00:35

of the market forces are happening on I

play00:39

don't know if you saw the SAP stand it's

play00:41

a little small stand at the corner in

play00:43

the back one or two people go say hi to

play00:45

them they feel a bit lonely so why do

play00:49

they feel lonely so what are some of the

play00:51

things that are happening um at least a

play00:54

little bit in the market conditions um

play00:56

as we're coming out of the covid area

play00:59

and the recovery is that there's a lot

play01:01

of um companies that are U not only

play01:05

merging right and Acquisitions but they

play01:07

have to merge more quickly and they have

play01:09

to get these uh financial statements out

play01:12

as as quickly as possible but then

play01:14

there's also divestures and if we look

play01:16

at that um that has a huge implication

play01:20

as concerning uh data and how it works

play01:23

so take for example there's

play01:26

traditionally lots of um companies out

play01:28

there that have made mergers after

play01:30

mergers after mergers and you have 30 or

play01:32

40 companies that are into one and they

play01:35

all have SAP systems but they're on a

play01:37

different blueprint how do you bring

play01:38

that financial data together uh easily

play01:41

mix it enrich it and whatnot um second

play01:45

point is BW they've been phasing it out

play01:47

well they've been phasing it out since

play01:49

like

play01:50

2012 um Hassel plaer it's like BW is

play01:54

dead and yet it keeps coming back and

play01:57

it's eventually getting phased out back

play01:59

in 2019 they said it we'll phase it out

play02:02

in 2024 and now it's 2027 that it's 2029

play02:06

and if you pay just a little bit more

play02:08

you can get a bit more but okay that

play02:12

makes people think a bit on the uh sap

play02:15

side but what customers are actually

play02:17

looking for is innovation uh sizing

play02:20

scalability

play02:22

stability and then the third part data

play02:26

collaboration po pi uh all these things

play02:30

you know this transactional pieces

play02:31

between uh company's great but it's only

play02:34

a megabyte you know in today's day and

play02:36

age uh we're not even talking about

play02:39

gigabytes anymore we're talking about

play02:41

terabytes and pedabytes and how many how

play02:44

many uh petabyte BW systems are out

play02:49

there it's compressed no okay so there's

play02:52

you know the the biggest every I've t I

play02:55

talk to the biggest companies on the

play02:57

planet and they all have the biggest sap

play02:58

system there there's literally 10 of

play03:01

them but um the largest BW system I've

play03:05

heard of uh it's not even running on a

play03:07

Hannah system but it's something around

play03:10

250 it's running uh IBM db2 somewhere in

play03:13

Canada but it's around 300 300

play03:17

terabytes any guess what the largest

play03:19

snowflake customer

play03:21

is how big it

play03:23

is 40 pedabytes

play03:26

what's that even mean it's huge so if

play03:30

you if you take if you take the largest

play03:32

BW system on Hannah which is about 120

play03:37

terab it's 277 times bigger than that so

play03:41

that's comparing Venus to the Sun so

play03:45

it's just literally a different Market a

play03:48

different way of working with that we

play03:51

you know we call it Big Data and then we

play03:52

call it data and then we call it Big

play03:54

Data again but being able to collaborate

play03:56

that means sharing pedabytes uh Gab

play04:00

terabytes pedabytes of data is super uh

play04:02

important in your supply chain uh with

play04:05

customers and without having to push

play04:07

this data back and forth across the

play04:10

networks so what are some of the

play04:12

challenges um on the Erp side again if

play04:15

you have 40 50 erps the some of the

play04:17

biggest companies on the planet uh

play04:19

getting that data together there's

play04:21

either you spend 15 years getting one

play04:23

blueprint onto that for example the

play04:26

globe project at nestate or you do uh

play04:29

that that consolidation piece either on

play04:31

the BW side or in a system like

play04:33

snowflake uh and you bring it together

play04:35

and you report on it the second part is

play04:38

being able to analyze that data it's

play04:40

fantastic you know if it's coming off an

play04:42

sap system why not use sap if you have a

play04:45

100%

play04:47

sap there's not a lot of value add on

play04:50

the snowflake side on the other hand if

play04:51

you have nonsp systems and you have

play04:54

semi-structured data or unstructured

play04:56

data and you want to be able to bring it

play04:58

in just e omally it's not it's doesn't

play05:01

make any sense to do this into a Hannah

play05:04

system you want something that can store

play05:06

that data that can govern the data that

play05:07

can centralize the data where you can

play05:09

access the data and you only pay for

play05:12

what you use um the third part um okay

play05:15

we you know expensive complex difficult

play05:17

to support but hey we you know we we all

play05:20

made our you know a lot of uh how do you

play05:22

say supported my family for a long time

play05:24

on that um but some of these challenges

play05:27

if we get down to the business side of

play05:29

things what kind of questions are we

play05:30

trying to answer we've been trying to

play05:31

answer these questions for 20 years but

play05:34

now we're talking about what's the most

play05:35

economical way to answer it what's the

play05:37

most um effective way of answering that

play05:40

what if I want to enrich that data what

play05:41

if I want to get that AI uh ml piece I

play05:44

have to say that at least 40 times or I

play05:47

get kicked off right AI ml um Global

play05:49

spend improving production sales Trends

play05:52

external data supplier variant uh

play05:55

inventory availability uh take an

play05:57

example of a company in in Germany tea

play06:01

they say okay we're going to take our

play06:02

sap data and we're going to take our uh

play06:05

shop floor data we're going to take that

play06:07

conveyor belt data and say hey um if

play06:10

there's a Slowdown in production or

play06:12

something stops is because I now combine

play06:15

that data in Snowflake I can tell oh

play06:18

this is com this is connected to this

play06:20

customer and I can send them a message

play06:22

right

play06:23

away um typical sap data architecture uh

play06:28

I like the slide I built no I didn't buy

play06:30

I didn't build the slide but having

play06:32

worked it right you have your

play06:34

transactional system and then you need

play06:35

to bring that into um you know

play06:38

traditionally as extractors and then

play06:40

bring that into uh for 20 years it was

play06:43

BW and then it's like oh wait we have to

play06:46

move into the cloud um oh wait we need

play06:49

to be in memory so we now you have a

play06:52

choice of BW or now I have a choice of

play06:53

Hannah okay I'll go BW oh wait I can't

play06:56

fit everything in so I need to use um

play07:00

hot warm cold storage and so I have the

play07:03

BW piece I have the Hannah piece I have

play07:05

the cbase piece I have the hadu lake

play07:08

piece and somehow it's all managed and

play07:11

work and somebody's going to partition

play07:12

it and somebody's going to push back and

play07:15

forth the data and I'm eventually going

play07:17

to get the data and report on it um and

play07:21

then just to add to that it's like okay

play07:25

we're moving to the cloud now that I've

play07:27

got the cloud um you want want your BW

play07:30

data into uh sap system well okay let's

play07:35

you need a another BW BW bridge and then

play07:38

then bring it

play07:38

into the inmemory reporting which is on

play07:41

Hannah which again has certain

play07:43

restrictions

play07:45

um

play07:46

no if we move and for example take

play07:49

what's the value at of of having

play07:51

snowflake for example surprise um one is

play07:56

that from any source of data any type

play07:59

volume variety velocity of data uh you

play08:03

can bring that into snowflake as a first

play08:05

class citizen it's very easy so

play08:07

traditionally what did we do we have

play08:09

extraction uh you wouldd have to

play08:10

transform that it have to be structured

play08:12

you have a lot of logic in your ETL

play08:14

process uh and then you land it into

play08:16

your Target and then if you have to

play08:18

change it you have to go back

play08:19

dependencies Etc um with modern data

play08:23

architecture right you extract the data

play08:26

as raw as possible and land it into you

play08:29

oh data Lake well snowflake is a

play08:31

combination it does both the data Lake

play08:33

and does the data warehousing and it's

play08:35

been doing it for 10 years now it's not

play08:39

you know it's been it's one of the first

play08:40

in the market to to do both that so one

play08:43

increase Enterprise

play08:45

visibility the second thing is that now

play08:48

hey I don't have to have siloed systems

play08:51

for my it data transactional data and my

play08:53

OT data right OT data shop floor sensor

play08:57

this can be 100 times or a th thousand

play08:59

times more data than your transactional

play09:03

systems and so you need to have an

play09:05

economical way of bringing that across

play09:07

and bble to store it and distort it into

play09:09

one place snowflake is a great place um

play09:11

deliver value faster how do you do that

play09:14

if you can land that data in one place

play09:16

and uh conform it and model it and

play09:18

report on it um because it's in one

play09:21

place it's easier but secondly there's

play09:23

just basic functions in Snowflake we

play09:25

call zero copy clone um we couldn't do

play09:27

this before so say I have a 100 terte uh

play09:31

system and I need to do some I I want to

play09:34

stress test it well I'm not going to

play09:35

stress test my production system so we

play09:38

have something called zero copy clone it

play09:40

makes a uh a zero copy clone so it

play09:43

doesn't take any more space than 100

play09:44

terabytes I can run different compute

play09:46

against it I can stress test all my data

play09:48

inside that and then go that and I can

play09:50

also zero copy clone it into Dev and

play09:52

test and things like that so instead of

play09:54

maintaining four systems right Dev test

play09:58

sandbox production

play09:59

I can generate and Destroy environments

play10:03

uh at

play10:05

will what are some of the things that we

play10:07

do on the business improvement process

play10:10

one enhanced forecasting because you can

play10:12

bring all the different data pieces um

play10:15

you can bring you know tax information

play10:17

it information that to bring it across

play10:19

uh sourcing supply chain risk um how

play10:22

many times do has your system your

play10:24

analytics systems fallen over end a

play10:26

month end a year as I I have zero

play10:29

visibility it is extremely hard to make

play10:32

a snowflake system fall over why because

play10:35

it's distributed because you have

play10:36

different we call them virtual

play10:37

warehouses for the computation we have

play10:40

the data stored in triplicate uh in the

play10:43

storage piece and and and whatnot so

play10:46

there's a a huge amount of redundancy

play10:48

and resiliency that's built into it for

play10:51

if you want a high availability in in

play10:53

sap system it only costs twice as much

play10:56

because you have to have two systems

play10:57

right snowflake is just built in

play11:00

um making right uh machine learning

play11:03

we're coming out with the co-pilot and

play11:05

whatnot uh delivery custom fulfillment

play11:08

uh you know use use that transactional

play11:10

data but be able to mix it with the

play11:11

other pieces uh and services being able

play11:14

to uh detect on your iot side on your

play11:17

sensor data uh it just enables you to

play11:19

react more quickly as opposed to right

play11:21

traditionally having these Silo

play11:25

systems who are some of the partners um

play11:27

that we work with right I I think one of

play11:29

the the fundamental uh key value pieces

play11:33

for snowflake is the ecosystem of

play11:36

partners that we have and because we

play11:38

have a cloud um based system is that any

play11:42

s snowflake customer can uh share data

play11:46

with any other customer and this can

play11:48

allow you in Upstream Downstream supply

play11:51

chain uh business partners to be able to

play11:54

share not megabytes of data but

play11:56

pedabytes of data this is super

play11:58

interesting on the master data side uhi

play12:01

spare spare parts for example uh Cold

play12:04

Storage these kinds of things um and so

play12:08

on the manufacturing piece we have a lot

play12:09

of c um a lot of uh snowflake Partners

play12:13

who are sharing their data either for

play12:15

free or as part of the marketplace and

play12:18

it's also a great way if you want to

play12:19

monetize your data really easily you

play12:21

immediately have access to over 9,000

play12:24

other uh customers on the technology

play12:27

Partners you know a AWS we're the number

play12:29

one

play12:30

isv for AWS this is this is a massive

play12:35

thing for us and and a great thing for

play12:37

for AWS on that side but we also work on

play12:40

you know Microsoft and and Google um

play12:42

powered by snowflake you can we first

play12:45

started out on the data side but now we

play12:47

have something called native

play12:48

applications and and we have lots of

play12:50

companies that are building applications

play12:52

uh directly on snowflake first it was

play12:54

analytical loads um and now we're

play12:56

bringing in our htap or uh um oltp piece

play13:02

uh as well into Snowflake and then

play13:04

systems integrators right especially on

play13:06

something as complex as as sap the

play13:09

partners you're working with today uh

play13:11

they also work with

play13:14

snowflake what are some of the steps the

play13:16

two big technical use cases that we see

play13:20

is that these large companies that have

play13:22

AWS U sorry Hannah Enterprise data

play13:25

warehouse um right costs Millions you

play13:28

have you know Max maybe 60 terabytes 100

play13:30

terabytes on there for only you know 16

play13:33

20 million a year is that a terabyte in

play13:37

Snowflake is 23 23 per terabyte per

play13:41

month so it's it's literally hundreds of

play13:44

times cheaper and that's just on the

play13:46

storage piece of course there's a

play13:48

computation piece but we've seen uh

play13:50

reductions in costs uh for equivalent uh

play13:53

features and even right improved

play13:55

features 40 50% and we have uh customer

play13:58

references on that so one is just

play14:01

complete replace of of Hannah Enterprise

play14:03

data warehouse uh it's an anql database

play14:06

at its foundations uh snowflake is as

play14:08

well and the second piece is is BW it's

play14:11

always a it's always a harder play right

play14:13

because it's been around for a long time

play14:15

there's a lot of custom code in it um

play14:17

and there's also right planning

play14:19

consolidation etc etc is that what if

play14:23

you look at the sap road map though

play14:24

they're pushing that planning piece out

play14:26

of BW into sap Analytics cloud and

play14:29

they're pushing the consolidation piece

play14:31

into group reporting on the S4 piece so

play14:34

one it's already moving out of BW second

play14:37

piece is that you can save a lot of

play14:38

money reducing that size of BW uh moving

play14:42

the workloads into Snowflake and then uh

play14:45

for the other

play14:46

piece uh you know why not keep a bit bit

play14:49

of BW around everybody everybody loves a

play14:51

little bit of

play14:53

that um you you know what are the basics

play14:56

moving Erp data at sap to snowflake

play14:59

systematically everybody ask me how do

play15:01

you get data out of

play15:03

sap there's 12 interfaces and we will go

play15:05

through each one no sorry um

play15:09

so there's there's some Partners there

play15:12

some Partners in the room that have been

play15:14

doing this for years and years and years

play15:17

uh sap silver partner uh they know how

play15:20

to get the data out they know how to get

play15:22

um what about transparent tables yes

play15:24

pool tables yes cluster table all that

play15:26

stuff okay that what about the security

play15:28

yes of course

play15:29

uh and how do I get that in real time

play15:31

yes uh you can get that out trigger

play15:34

based batch based micro batch um it's

play15:37

all there it's been done it's been done

play15:39

thousands of times um and then

play15:41

acquisition so there's uh partners that

play15:44

push data directly into snowflake tables

play15:46

there's other ones that use files um

play15:49

depends on how fast you want it depends

play15:51

on some of the economics transformation

play15:54

some

play15:55

people snowflake is a is a fully

play15:58

functional not only on the SQL feature

play16:01

side but python Scala Java all these all

play16:05

these and the and the libraries that are

play16:07

behind it you can use all these

play16:09

different tools they're at your disposal

play16:11

inside of that uh centralized governed

play16:14

snowflake um uh ecosystem platform as it

play16:18

were uh reporting analysis right you can

play16:21

we're model agnostic I was talking to

play16:24

customers using Data Vault bring across

play16:26

30 different BW systems into to that

play16:29

other ones that saying I'm going to use

play16:31

BW it's already modeled I use that as a

play16:33

source and so I have my 65 dimensions

play16:35

and my three fact tables for my Global

play16:37

spend analysis and and it's it's there

play16:40

already uh reporting say obbc

play16:43

jdbc uh again if you want to use python

play16:46

or ml AI all that all those interfaces

play16:49

are there uh and then at the end of

play16:51

course right you want to be able to

play16:52

eventually turn off uh some of other

play16:55

stuff and get those cost savings we have

play16:57

a uh a breako right you're going to have

play16:59

both systems running at the same time so

play17:01

that's going to cost a bit more but then

play17:03

you have a massive uh decrease in spend

play17:05

on the snowflake side and as you turn

play17:07

off that BW

play17:10

side um some of the uh uh partners that

play17:15

we work with I'll just talk about um on

play17:18

the acquisition side there's S&P glue

play17:20

can you raise your hands so these two

play17:23

guys S&P glue us be data data glue uh

play17:28

worked really closely with us over the

play17:29

past year we came out with uh it's

play17:32

called a streaming SDK and so previous

play17:35

at snowflake you would have to batch

play17:37

load things staging area bring it across

play17:40

they worked with us and so you can now

play17:42

stream data directly from an sap system

play17:45

uh directly into a snowflake system it

play17:47

is the most direct path on the market

play17:51

for the data is is these guys here um

play17:53

Informatica they came in a bit uh the

play17:56

other ones might work with some other uh

play17:58

cases and they're absolutely

play18:03

available um if we take a look at some

play18:05

of the uh reference

play18:07

architectures uh pretty simple there's

play18:10

you know you need to have some kind of

play18:12

uh tool between SAP systems and

play18:16

snowflake uh and beyond that is like

play18:20

what's the interest of snowflake is like

play18:21

take over the data Lake component data

play18:24

engineering component uh ml component AI

play18:27

component um the other other workloads

play18:30

we also have you know um security logs

play18:32

and and whatnot but you land it uh you

play18:35

can transform it within snowflake you

play18:37

can use a tool of choice or you can use

play18:40

uh Native features that we have in

play18:41

Snowflake we have something Dynamic

play18:43

tables say you have you use an S&P glue

play18:45

you land that you have that PSA that raw

play18:48

layer and then a dyamic tables you can

play18:50

do any kind of arbitrary uh sequels that

play18:54

combine these five tables uh to have

play18:56

this uh uh reference table at end and we

play18:59

manage the Delta we manage uh the

play19:02

loading behind it

play19:07

automatically some of the replication

play19:09

tools some of the uh uh and we'll leave

play19:12

this as as reference because it gets a

play19:13

bit a bit complicated on what can you

play19:17

hit against uh what does things natively

play19:21

what does things at the database level

play19:23

and whatnot but and um and this is just

play19:26

on the sap side right what are the what

play19:28

the sap supported tools uh or sap tools

play19:33

that have um certified snowflake

play19:35

connectors so data services

play19:39

um data

play19:41

services and so they can use also these

play19:44

other tools that have that and bring it

play19:47

across uh third party tools is it a Ab

play19:50

App application that's putting pushing

play19:51

data out like the glue is it a some

play19:54

agent that's sitting outside of that um

play19:57

is it hitting against database logs is

play20:00

it how are you dealing with the sap

play20:04

licensing if we take a look at the AWS

play20:09

so this is a a general reference

play20:10

architecture that I built up with the uh

play20:13

AWS Architects and I think it's one of

play20:15

the especially for manufacturing one of

play20:17

the most compelling architectures out

play20:19

there why is because you can bring in

play20:21

not only that it data transactional

play20:24

Oracle sap or whatnot but also that OT

play20:27

data in one place an economical

play20:31

efficient effective way of bringing it

play20:34

across um

play20:36

and it goes through a lot of the

play20:38

different uh pieces that we have um

play20:42

maybe a bit of you know snowflake 101 is

play20:45

that we land the data once it's in

play20:47

triplicate for uh resiliency and then we

play20:51

have compute clusters on top of that and

play20:53

you only pay by the second say I run a

play20:55

query uh we snowflake have a hot pool of

play20:59

compute clusters that are running um so

play21:01

it attaches in a second runs the query

play21:04

and then it can turn off and so you're

play21:05

only paying for that number of seconds

play21:08

that that compute cluster is running

play21:09

plus you know average storage cost you

play21:11

look at Legacy systems transactional

play21:14

systems even Cloud solutions that are

play21:16

supplied other things it takes the other

play21:18

Solutions right 15 minutes to start then

play21:21

I can run my query and then I don't want

play21:23

to turn it off to this cash you're

play21:25

paying for that all the time that's not

play21:27

the case with snowfl like turn it on off

play21:30

it's like a utility it's like

play21:34

electricity AB they've been my C so I've

play21:37

worked for snowflake for for four years

play21:39

worked sap before that for 18 years uh

play21:42

ABB has been my customer for 10

play21:44

years we'll do the math um they have 40

play21:50

uh erps and you think okay it's all sap

play21:53

uh just bring it across and it's easy to

play21:55

report no because it's 40 different

play21:57

blueprints

play21:59

and so they spent um literally you know8

play22:02

years uh building up a semantic layer on

play22:05

the BW side because all the fields

play22:07

weren't same matching and things like

play22:09

that a lot of uh manual effort there and

play22:12

now you know now it's easy enough the

play22:15

IPS are easy enough to shift over uh to

play22:18

snowflake bring together have a modern

play22:20

uh platform that's scalable not to 60

play22:24

terabytes but to hundreds of terabytes

play22:26

you know to pedabytes um um some of the

play22:29

savings that they have right 200 200

play22:32

million savings annually on the U

play22:35

inventory purchases uh Revenue growth

play22:37

they have to have this to support it how

play22:40

else can you go from a couple hundred

play22:42

analytics users to thousands and

play22:45

thousands of analytic

play22:47

users it's snowflake it it can scale it

play22:50

scales horizontally it scale vertically

play22:52

within seconds it's not a it's not a

play22:55

question like oh I need to order uh

play22:57

another machine it will come in weeks I

play22:59

need to transfer the No it's it's in

play23:01

seconds it's like make it make it bigger

play23:03

make it extra large make it small

play23:06

um the next one uh near and dear to my

play23:10

heart seens anybody here seens no no so

play23:14

it's a and some two people from Seaman

play23:16

are right here um right uh absolutely

play23:20

huge Hannah Enterprise data warehouse

play23:22

deployment and within months uh they

play23:25

replaced that with snowflake talking

play23:27

over 50 uh

play23:29

erps uh they're using a combination of

play23:32

uh sap technology and S&P glue to get

play23:35

the data across so SLT plus SNP uh land

play23:39

it and snowflake all the Transformations

play23:41

now they're able to serve up tens of

play23:42

thousands of users uh they're able they

play23:45

build an application to uh automatically

play23:48

generate and deploy snowflake

play23:50

environments for for their users and

play23:52

this is it automating this creation

play23:55

which unheard of um so it reduces supply

play23:58

chain risk

play23:59

uh improved cost and efficiency gains um

play24:02

secure self-service tools which we just

play24:06

mentioned so how do you get

play24:09

started um do you want to no okay so one

play24:15

of the things to go is like what's

play24:17

what's your status quo what for example

play24:20

if you're using BW what are you not

play24:23

using and wouldn't it be nice if you're

play24:24

not using 30% of it that you're not

play24:26

going to uh migrate it across

play24:29

and so uh S&P for example gives a uh

play24:32

we're going to change the name to

play24:33

Fitness tal or like you know let let BW

play24:35

die nicely but um maybe you won't but

play24:41

the uh um you install it runs four or

play24:44

five weeks you see what's what's being

play24:47

used what's not being used and then you

play24:48

can decide uh hey if we're not using

play24:50

that we can turn it off and maybe BW

play24:52

will live a little bit longer or hey

play24:54

these are the things that are super

play24:55

important let's bring those across um

play24:58

thank you I'll open up to questions we

play24:59

have a couple minutes

Rate This

5.0 / 5 (0 votes)

Related Tags
SnowflakeSAPデータ分析製造業市場動向統合スケーラビリティクラウドデジタルトランスフォーメーション効率化技術革新
Do you need a summary in English?