SAP – DB2 V10.5 mit BLU Acceleration · PDF fileDB2 Optimierungen für SAP COPA und...

23
SAP – DB2 V10.5 mit BLU Acceleration Die neue IBM In-Memory Technologie eröffnet neue Möglichkeiten für Ihr Business

Transcript of SAP – DB2 V10.5 mit BLU Acceleration · PDF fileDB2 Optimierungen für SAP COPA und...

SAP – DB2 V10.5 mit BLU AccelerationDie neue IBM In-Memory Technologie eröffnet neue Möglichkeiten für Ihr

Business

SAP auf DB2 Entwicklungs-Team

SAP - DB2 Integration Center

� Integration von neuem DB2 Code mit existierenden SAP Releases

� DB2 QA für jeden neuen DB2 code level mit SAP Anwendungen weit vor IBM GA

� SAP Development Support

� Zusammenarbeit mit DB2 Service

� Gemeinsames iBM und SAP Team

SAP - DB2 Development

� Entwicklung von SAP Code

� Entwicklung von DB2 Code für SAP-spezifische Funktionen

� SAP Development Support

� Zusammenarbeit mit DB2 Service

� Gemeinsames IBM and SAP Team

SAP auf DB2 ist ein vollständig integriertes Produkt

SAP auf DB2 ist ein voll integriertes Produkt

� Integrierte Installation der DB2 Software mit SAP install *

� Integrierter Hochverfügbarkeits Setup mit SAP install *

� One-step SAP-DB2 Konfiguration: DB2_WORKLOAD=SAP

� Komplette DB2 Administration und Monitoring mit SAP DBA Cockpit

One-stop Support

� Alle Kunden erhalten "One-stop Support" durch SAP – nur ein einziger Kontakt

� Gemeinsames IBM und SAP Support Team

Synchronisierte Wartungszyklen

� IBM DB2 folgt der SAP 7+2 Wartungsstrategie

*as of SAP NetWeaver 7.0 SR3 and SAP Applications based on NW 7.0 SR3

DB2 Optimiert für SAP - Roadmap

Eine Auswahl der “SAP on DB2 Technology Highlights”

6

DB2 Optimierungen für SAP COPA und andere SAP Anwendungen

SAP COPA Profitabilitäts-Kalkulation• Komplexe SQL Queries mit grosser Zahl aggregierter Zeilen (ähnlich zu BW)� Kandidat für Column-store und In-memory Technologie

COPA: Nutzung DB2 10.5 Parallel Processing (ohne BLU)• GLEICHE Hardware !!!• DB2 parallel degree: 1 -> 8 � bis zu 4x schneller• DB2 parallel degree: 1 -> 16 � bis zu 7x schneller

COPA: Resulte mit DB2 9.7• GLEICHE Hardware !!!• bis zu 3.8x schneller mit DB2 parallel processing• im Schnitt 1.8x schneller mit DB2 parallel processing

SAP Bank Analyzer / SAP Retail• Study on Horizontal Scalability of a Typical SAP Bank Analyzer Scenario on IBM DB2 10.1 pureScale and POWER7

(http://scn.sap.com/docs/DOC-43486)• SAP Enterprise Data Warehouse for Point of Sales Data Optimized for IBM DB2 for Linux, UNIX, and Windows on IBM

Power Systems (http://scn.sap.com/docs/DOC-14457)

SAP COPA Profitability Calculation

0,00

2,00

4,00

6,00

8,00

10,00

12,00

1 2 3 4 5 6 7 8 9

hours

Without parallelprocessing, run time (h)

With parallelprocessing, run time (h)

2.7 Average1.5 Average

7

� Separate SAP BW Online Daten und Near-Line Storage (NLS) Daten

� Vollkommen transparent für SAP BW Anwendung – keine Änderung der Anwendung notwendig !

� Kleine, schnelle Online Datenbank für häufig genutzte Daten

� Grosse (und langsamere) Near-Line Datenbank für „archivierte“ Daten

� DB2 Alleinstellungsmerkmal

DB2 Nearline-Storage für SAP-BW – niedrigere Kosten, bessere Online-Performance

Transparent AccessTransparent Access

SAP- BW Online Database

high performance

storage

DBMSTREX

DB Interface

Layer

Relational DB

Interface

BI Data Manager

General NLS

Interface

NLS / DB2 LUW

Interface

BI OLAP

DBMSTREX

SAP NetWeaver BW

DB Interface Layer

Relational DB Interface

General Near-Line Interface

Near-Line PartnerInterface

BW OLAP

Near-Line DB2 Database

low cost storage

DB2 10.5 Optimiert für SAP Anwendungen Höchste Performance & niedrige Kosten

Verfügbarkeit

DB2 10.5 ist seit Juni 2013 verfügbarDB2 10.5 für SAP ist seit August 2013 verfügbarDB2 10.5 für SAP mit BLU Acceleration erwartet bis Ende 2013

Lizenz

OEM: BLU wird Bestandteil der SAP DB2 ASL Lizenz sobald zertifiziertDirekt: DB2 Advanced Enterprise Server Edition beinhaltet BLU (ohne zusätzliche Kosten)

"IBM is working closely with SAP to certify DB2 10.5 in similar time frame as previous major

releases. This usually occurs about 8 weeks, give or take, after we GA, so, assume late August for the certification statement from SAP. As with any release, this includes evaluation and

exploitation of all features in this release where appropriate (including BLU acceleration). We

would be happy to help arrange and participate in a joint meeting/call with you, SAP and IBM to discuss SAP's plans further.“

- Torsten Ziegler, Development Manager SAP DB2 porting Team, SAP

• 8-25x schneller bei Reports und Analysen1

bis zu 1000x schneller bei manchen Queries2

• 10x Storageeinsparung3

• Nahtlos integriert in neue DB2 Version 10.5 für einfache “out of the box” Nutzung auf bestehender Infrastruktur

BLU Acceleration

1 Based on internal IBM testing of sample analytic workloads comparing queries accessing row-based tables on DB2 10.1 vs. columnar tables on DB2 10.5. Performance improvement figures are cumulative of all queries in the workload. Individual results will vary depending on individual workloads, configurations and conditions.

2 Based on internal IBM tests of pure analytic workloads comparing queries accessing row-based tables on DB2 10.1 vs. columnar tables on DB2 10.5. Results not typical. Individual results will vary depending on individual workloads, configurations and conditions, including size and content of the table, and number of elements being queried from a given table.

3 Client-reported testing results in DB2 10.5 early release program. Individual results will vary depending on individual workloads, configurations and conditions, including table size and content.

Eine neue Generation von Daten-Management Innovationen

BLU AccelerationBLU Acceleration

Instructions Data

Results

C1 C2 C3 C4 C5 C6 C7 C8C1 C2 C3 C4 C5 C6 C7 C8

Dynamic In-Memory Spaltenorientierte In-Memory Verarbeitung mit dynamischer Auslagerung nicht genutzter Daten auf Storage

Actionable CompressionEinzigarte Datenkomprimierung mit Beibehaltung der Sortierreihenfolge ermöglicht Nutzung der Daten ohne Dekomprimierung

Parallel Vector ProcessingMulti-core und SIMD Parallelverarbeitung(Single Instruction Multiple Data)

Data SkippingIrrelevante Daten werden bei der Verarbeitung übersprungen

EncodedEncodedEncodedEncoded

Super Fast, Super Easy—Create, Load and Go!Keine Indices, keine Aggregate, Kein Tuning, Keine SQL- oder Schema-Änderungen

Warum ist BLU Acceleration einzigartig ?Unerreichte IBM Forschungs- und Entwicklungs-InnovationenWarum ist BLU Acceleration einzigartig ?Unerreichte IBM Forschungs- und Entwicklungs-Innovationen

11

BLU Acceleration Illustration10TB Query in Sekunden oder schnellerBLU Acceleration Illustration10TB Query in Sekunden oder schneller

Das System: 32 cores, 1TB memory, 10TB Tabelle mit 100 Spalten und Daten über 10 Jahre

Die Query: Wie viele Abschlüsse hatten wir in 2010? SELECT COUNT(*) from MYTABLE where YEAR = ‘2010’

Das Ergebnis: In Sekunden oder weniger weil jede CPU nur 8 MB Daten untersuchen muss

10TB data

Actionable Compressionreduziert Daten auf 1TB

In-memory

Parallel Processing32MB linearer Scan auf jedem Core via

Vector Processing Scan so schnell wie bei

8MB durch SIMD Ergebnis in Sekunden oder

weniger

Column Processingreduziert zu 10GB

Data SkippingReduziert zu 1GB

DATADATADATADATA

DATADATA

DATADATADATADATA

DATADATA

DATADATADATADATA

DATADATA

DATADATA DATADATA DATADATA

DATADATADATADATA DATADATA

DB2 BLU - Seamless Integration into DB2

Built seamlessly into DB2 – integration and coexistence

Column-organized tables can coexist with existing, traditional tablesSame schema, same storage, same memory

Same SQL, language interfaces, and administration

Column-organized tables or combinations of column-organized and row-organized tables can be accessed within the same SQL statement

Dramatic simplification – Just “Load and Go”

Faster deploymentFewer database objects required to achieve same outcome

Requires less ongoing management Due to its optimized query processing and fewer database objects required

Simple migrationConversion from traditional row table to BLU Acceleration is easy

Users only notice speed-ups; DBAs only notice less work!

13

� Integration into SAP BW Workbench� SAP ABAP Dictionary extension to support BLU tables as new table type� BLU conversion of existing BW objects� DBA Cockpit: Support of new performance metrics for BLU tables� Configuration for BLU feature is fully compatible with SAP settings

- DB2_WORKLOAD=SAP

DB2 BLU Integration into SAP BW

14

• Report SAP_CDE_CONVERSION_DB6

– Non-BLU -> BLU in online mode

– BLU -> BLU / Non-BLU in read-only mode

• Select an InfoCube or all InfoCubes

• "Get Dependent Tables" determines the tables belonging to the BW Object

DB2 BLU Conversion of existing BW Objects

15

• SAP NetWeaver BW 7.00 and higher

• Support expected to start with DB2 10.5 FP1 (End of 2013)

• DB2 10.5 BLU extensions are delivered with SAP BW support packages

Planned SAP BW Adoption for DB2 10.5 BLU Feature

Common

8x-25x

improvement

Common

8x-25x

improvement

How fast is it? Results from the DB2 10.5 2nd Alpha Customer Tests

Workload Speedup over DB2 10.1

Large Financial Services Company 46.8x

Global ISV Mart Workload 37.4x

Analytics Reporting Vendor 13.0x

Global Retailer 6.1x

Large European Bank 5.6x

Internal Benchmark Test 3.0x

17

DB2 = ONE 4 ALL

Average diaglog response time 0,2 - 0,8 sec

Average diaglog response time 0,4 – 2 sec

Transactional Analytical Near-line Storage

SAP Business-Suite,Industry Solutions

OLTP workload

SAP BW Big Data

OLAP workload

NLS for BW

Near-line Storage

DB2 LUW DB2 LUW DB2 LUW

18

� 4 POWER servers

DB2 - Tailored Performance for SAP Customerscompared with non-virtualised and non-consolidated solution

Customer runs DB2 on POWER/AIX

- 180 systems, 48 production

- 26 HA (LPM*) + 26 DR (PowerHA)

- 2 data centers

Implementation w/o virt. & consol.- 180 systems, 48 production- 26 HA + 26 DR clusters- 2 BIGGER or more data centers

* LPM - AIX live partition mobility

� 101-232 servers

- 48 servers for production- 52 servers for HA+DR clusters- up to 48 servers for test/QA- up to 48 servers for dev- up to 36 servers for rest

19

Compression and Storage Savings

• Often compression and storage savings are wrong determineddatabase size consists of ALL data stored on storage (e.g not only tables)

• Comparing in-memory databases and on-disk databases should be done on the right level

On-disk database RAM

On-disk database storage

In-memory DB RAM

In-memory database storage

© 2013 IBM Corporation

Cost Comparison for DB2 on Power vs. SAP HANA on Intel

EXAMPLE: 10TB raw active user data

• DB2 10.5 BLU recommendation for 10TB of data is 500GB - 1TB of memory

• SAP recommendation for 10TB of data is 4TB - 5TB of memory

• DB2 10.5 BLU storage requirements for 10TB of SAP data is 1.4TB storage

• SAP storage requirements are = 4 * memory required = 16TB – 20TB

• DB2 10.5 BLU CPU requirements for 10TB is 32 cores of Power7+

• SAP HANA CPU requirements for same database is 320 cores of Intel

• Costs

• DB2 10.5 on 32core 770+ server (1TB DRAM & 2TB SSD storage)

• Software costs = DB2 AESE = $94K/TB = $188,000 (2TB of compressed data)

• Hardware costs = $878,884

• Total = $1,066,884 (including 1st year support)

• SAP HANA on 9 40-core IBM x3590 servers (512GB DRAM & 2.5TB storage each)

• Software costs = SAP HANA Enterprise Edition for 2TB compressed data = $6,106,000

• Hardware costs = $$1,450,000

• Total = $7,556,000 + $1,343,320 (1st year S&S) = $8,899,320

SAP HANA is 834% more expensive than DB2 10.5 BLU on Power

Was unterscheidet DB2-BLU von SAP-HANA ?

DB2- BLU

• Wettbewerbsvorteile durch neue und schnellere Prozesse

• Investitionssicherheit durch SAP-IBM- Kooperation

• Standard- Hardware

• volle Flexibiltät bei der Plattform –Auswahl (inkl. IBM-Power-Plattform)

• Standard DB2-Betriebsaufwendungen

• Lizenzkosten in DB2 AESE inkl.

SAP- HANA

• Wettbewerbsvorteile durch neue und schnellere Prozesse

• Investitionssicherheitdurch „SAP- Komplettlösung“

• spezielle Appliance- Hardware

• reduzierte Plattform- Flexibiltät (nur freigegebene x86-Systeme)

• spezielle Betriebsaufwendungen

• extra SAP- Lizenzzahlungen

• Storage- / Server- Kosten sinken • Storage- / Server- Kosten steigen

“When we compared the performance of column-organized tables in DB2 to our traditional row-organized tables, we found that, on average, our analytic queries were running 74x faster when using BLU Acceleration.” - Kent Collins, Database Solutions Architect, BNSF Railway

“What was really impressive is the fact that we could get significantly better performance with DB210.5 using BLU Acceleration without having to create indexes or aggregates on any of the tables. That is going to save us a lot of time when designing and tuning our workloads.” - Kent Collins, Database Solutions Architect, BNSF Railway

First Customer Quotes on DB2 10.5 Significantly Less Storage, Better Performance

“When I converted one of our schemas into DB2 10.5 with BLU Acceleration tables, the analytical query set ran 4-15x faster.” -Andrew Juarez, Lead SAP Basis and DBA

“10x. That's how much smaller our tables are with BLU Acceleration. Moreover, I don't have to create indexes or aggregates, or partition the data, among other things. When I take that into account in our mixed table-type environment, that number becomes 10-25x.” -Andrew Juarez, Lead SAP Basis and DBA

THANK YOU

Super analyticsSuper easy

DB2 with BLU Acceleration