LSA++ for BW on HANA - a248.g.akamai.neta248.g.akamai.net/n/248/420835/8bd646f3875ba50bbce... · developments are subject to change and may be changed by SAP at any time for ... Operational

SAP HANA as Driver of EDW Evolution:

LSA++ (Layered Scalable Architecture) for

BW on SAP HANA

Juergen Haupt, CSA Architect

SAP AG

August 2012

© 2012 SAP AG. All rights reserved. 2

Legal disclaimer

This presentation is not subject to your license agreement or any other agreement

with SAP. SAP has no obligation to pursue any course of business outlined in this

presentation or to develop or release any functionality mentioned in this

presentation. This presentation and SAP's strategy and possible future

developments are subject to change and may be changed by SAP at any time for

any reason without notice. This document is provided without a warranty of any

kind, either express or implied, including but not limited to, the implied warranties of

merchantability, fitness for a particular purpose, or non-infringement. SAP assumes

no responsibility for errors or omissions in this document, except if such damages

were caused by SAP intentionally or grossly negligent.


LSA++ The Holistic Picture on Structured Data for BI

LSA++ Principal Layers & BI Value Areas – Comparison to LSA

LSA++ Streamlined EDW & Architected Data Marts

LSA++ Open Operational Data Store Layer - Operational BI

LSA++ Virtual Data Mart Layer

LSA++ Agile BI


BW EDW LSA Grid

Consistent EDW Architecture – Consistent BI & Reporting

BW EDW Standard LSA Layers

Layer and naming qualifier

BW EDW LSA Data Domains

(standard Partitioning) Example:

C D B F E A

Standard Logical Partitions (Domains) and naming qualifier

Naming (e.g. DSO)

‒ Layer: byte 1

‒ Content Area: byte 2 to 5

‒ Sequence number: byte 6

‒ Domain: byte 7

‒ Sub-Partition: byte 8

Technical name

of DSO


EDW, Data Marts and Model Richness

EDW

data model

Business View Data Mart Model

source

data model

EDW

transform

Data Mart

transform

EDW Model Level

(Architected) Data Marts

Source Model Level

Data Marts

Business View Data Mart model

source

data model

Data Mart

transform


BW EDW with LSA is the accepted approach (not just for large companies)

guaranteeing standardized BI & Reporting on all organizational levels (local, regional,

global) on the same consistent, common data foundation (single version of truth)

BW EDW with LSA stands for:

High availability - 24x7

operational robustness

organizational independency

scalability

maintainability e.a.

The LSA is the corporate framework

enabling BW managing reliably

the entire data & meta data life cycle:

Data delivery (extraction, RDA)

Data modeling (assign InfoObjects)

Data staging

Authorizations

Solution delivery (transports)

BW Layered Scalable Architetcure - Value & Positioning

Consistent, High Availability EDW for Standardized BI














Open Operational Data Store Operational Extension

BW / Externally managed

LSA++ Holistic Framework

LSA++ Principal Persistent Layer

EDW Layers

Ag

ile

Da

ta M

art

La

ye

r

BW

Wo

rks

pa

ce

(La

ye

r) Flexible Consistent

EDW Core

Architected

Data Marts

LS

A+

+

Prin

cip

al L

aye

r

Dep

artm

en

tal

Ag

ile E

xte

ns

ion

Cen

tral A

gil

e

Exte

nsio

n



BI Flavours and LSA++ Principal Layers

Different BI flavours like EDW based BI and Operational BI request for different

services. LSA++ takes this into consideration structuring data by purpose/ context with

respect to BI requirements - this results in the principal layers:

Flexible Consistent EDW Core: EDW context data – Standardized template-based BI and Reporting on harmonized, consistent data

Operational Extension of the Core: Operational/ source context data – Operational / real time BI on source-level data

Agile Extension of the Core: Agile, ad hoc context data – Agile BI on all kind of data – single usage

LSA++ purifies the flexible consistent EDW Core offering dedicated services for

operational & agile purpose data (free the core from cholesterol)

This by itself increases overall LSA++ framework flexibility and manageability

LS

A+

+

Prin

cip

al L

aye

rs

LSA++

Operational

Extension

Operational,

source

context data

LSA++

Flexible

Consistent

Core

EDW context data

LSA++

Agile

Extension

Agile/ad hoc

context data



Principal Layers Characteristics & BI Solutions

BI solutions based on different LSA++ principal Layers have decisive

differences with respect to e.g.

• Data model

• Solution time-to-market

• Data Processing

• Ownership

• Consistency

• Stability & Robustness

• BI query result stability

• Revision capability.....

Data model: Deployment:


LSA++ Holistic Framework for BW on HANA

Flexible Consistent Data Framework Virtually Wrapped

BI on Streamlined EDW - Operational BI – Agile BI – Virtualization - Open BW Services

BW

Op

en

Se

rvic

es

: D

ata

Mo

de

l,

OL

AP

, A

uth

ori

za

tio

n,

Data

Ag

ing

BW Virtual Data Mart Layer

Architected Data Marts

Business Transformations

EDW Propagation

Layer

EDW Transformations

Externally managed BW managed

Open Operational Data Store

Agile

DMs/

BW

Work

space

Cen

tra

l B

I- P

rom

ote

To

Pro

du

cti

on

Ag

ile

BI -

ce

ntr

al/

de

pa

rtm

en

tal

LS

A+

+ fo

r BW

on

HA

NA

B

I on S

tream

lined E

DW

,

Opera

tional B

I, Agile

BI

Co

rpo

rate

Me

mo

ry

refe

ren

ce

BW Queries

EPM , BO BI, Apps







LSA++ Agile BI


LSA++ EDW & Architected Data Marts Value Scenarios

Reliable, managable, consistent data

Streamlined EDW & Architected Data Marts thru HANA-Optimized InfoProvider

Streamlined Architected Data Marts thru Virtual Data Marts

Streamlined EDW – Queryable EDW

Streamlined EDW thru Direct Data Provisioning into EDW Core

Real Time Master Data for Flexibility

Streamlined EDW & Architected Data Marts thru Re-modeling

– Flexible Master Data Modeling thru virtually joining semi-stable master data

– Large Master Data Modeling avoiding realignment thru compounding instead of

fact-table join

– Inventory Scenarios

– ......


LSA++ Layer Structure of Flexible Consistent EDW Core The LSA Heritage - Reliable, Managable, Consistent Data

LSA++ inherits the service definitions of LSA Layers that stands for reliability

& consistency:

EDW Transformation Layer (Link between source-model and EDW-model)

(EDW) Corporate Memory (Persistent )

(EDW) Propagation Layer (Persistent )

Business Transformation Layer (Link between EDW-model and Data Mart model)

Architected Data Mart Layer (Persistent )

These Layers define the consistent Core:


LSA++ for BW on HANA Value Scenario Streamlined EDW - EDW Propagation Layer

LSA++ EDW Propagation Layer using HANA Optimized DSOs means

Increased flexibility thru dramatic decrease of load and activation time

Increased modeling flexibility

Increased flexibility & value thru comprehensive data content offering of Propagation Layer

Relaxed volume considerations during design time (Domains/ semantical Partitioning)

Increased flexibility thru direct Querying on Propagation Layer (Streamlining Architected

Data Marts)

comprehensive

fast

dir

ect

Flexible, relaxed modeling


LSA++ for BW on HANA Value Scenario Streamlined Architected Data Marts

LSA++ Architected Data Mart Layer should use HANA Optimized InfoCubes

and/ or HANA Optimized DSOs (Business Transformation Layer) in case there

is a need for persistent Architected Data Mart (s. Virtual Data Marts)

LSA++ Architected Data Mart Layer using HANA Optimized InfoCubes means

Increased flexibility thru dramatic decrease of load time

Increased modeling flexibility

No multi-dimensional modeling

skills needed

Relaxed volume considerations

during design time

(-> Domains/ semantical

Partitioning)

fast

Flexible, relaxed modeling


1. Queries on DSOs (sid-enabled) show

similar performance compared to

queries on InfoCubes

2. BW Composite Provider feature

allows defining Virtual Data Marts

combining LSA++ EDW Core

InfoProviders

LSA++ Streamlined EDW Fundamental Changes in LSA++ with respect to LSA




EDW Propagation

Layer

EDW Transformations

Co

rpo

rate

Me

mo

ry

MultiProvider

Composite Provider


LSA++ Streamlined EDW Propagation Layer as Query Target

In publications we can read that the EDW in general is not query-able.

The reason for this statement is not that the EDW content is of no value on the

contrary – the reason is more that queries running on EDW tables on RDBMS will

most likely not come back (no secondary index = full table scan)

With LSA++ and BW EDW on HANA you can run queries on Propagator Layer

DSOs (side-enabled) with similar performance like on InfoCubes


LSA++ Streamlined Architected Data Marts Obsolete: InfoCubes as Accelerator on Business Transformation Layer DSOs


LSA++ Streamlined Architected Data Marts

CompositeProviders – Virtualization or Persistent Join ?

Scenario:

• Multiple DataStore Objects loading into a single InfoProvider

• Nowadays the relation between data (“join”) is modeled in BW transformation (routine) and /

or through updating (overwrite) loaded one target DataStore Object

• UNION Join in MultiProvider doesn’t correspond to reporting requirements

C Reporting Reporting

LSA

SAP NetWeaver BW

LSA++

SAP NetWeaver BW

powered by HANA

Composite

Provider


Decision Criteria on Using Composite Provider Architected Data Marts and the BW Virtual Data Mart Layer

Goal: Streamline Architectured Data Mart Layer

Granularity / Cardinality Architected Data Mart

vs. EDW Propagation Layer

Transformation / Join logic EDW Propagation Layer to

Architected Data Mart High/

same

low

no low complex

Investigate

Replace Data Mart

Keep Data Mart

Investigate


LSA++ for BW on HANA Value Scenario - Flexible InfoProvider Model

Composite Provider Joining Core Providers with Local Masterdata

C_COSTC C_ATTR_1 C_ATTR_2

1 A AA

2 B BB

3 C CC

G_COSTC G_ATTR_1 C_COSTC

1 X 1

2 Y 2

3 Z 3

U_COSTC U_ATTR_1 C_COSTC

1 O 1

2 P 2

3 Q 3

DATE C_COSTC C_AMOUNT

20120301 1 30

20120301 2 50

20120301 3 60

20120302 1 10

20120302 2 20

CompositeProvider

Germany

CompositeProvider

US

Corporate Master Data

InfoProvider

Corporate

German Master Data

US Master Data







LSA++ Agile BI





BW

Op

en

Se

rvic

es

: D

ata

Mo

de

l,

OL

AP

, A

uth

ori

za

tio

n,

Data

Ag

ing




EDW Propagation

Layer

EDW Transformations



Agile

DMs/

BW

Work

space

Ce

ntr

al B

I- P

rom

ote

To

Pro

du

cti

on

Ag

ile

BI -

ce

ntr

al/

de

pa

rtm

en

tal

LS

A+

+ fo

r BW

on

HA

NA

B

I on S

tream

lined E

DW

,

Opera

tional B

I, Agile

BI

Co

rpo

rate

Me

mo

ry

refe

ren

ce

BW Queries

EPM , BO BI, Apps


LSA++ Open ODS Layer Services -

Integration, Operational Data & EDW Services

The Open Operational Data Store Layer is the LSA++ Inbound Layer and as such

the home of source-level data delivered either directly to HANA or via BW.

The data are delivered to the Open ODS Layer via Extraction, real-time Replication

or File thus the Open ODS Layer inherits the field-level data models and values from

the sources (OLTP).

Core Services of the Open ODS Layer are:

Integration Services for data delivered directly to HANA or via BW

– Consume HANA Modeler schemas in BW and vice versa

– Transfer HANA Modeler managed data into BW managed area and vice versa

Operational Data Services

– Immediate querying on delivered data (Operational BI) - no staging to EDW Layers required

– Data Modeling on Open ODS Layer field-level Provider

EDW Services

– Open ODS Layer Provider as source for persistent EDW providers - data consistency

management (delta), data transfer (staging) and data transformation services for EDW

– Open ODS Layer Provider as virtual part of the EDW (if feasible from EDW standpoint) -

reduced redundancy, responsiveness, incremental EDW


LSA++ Open ODS Value Scenarios

Holistic picture of externally, HANA Modeler managed and BW managed data

Scalable Integration of HANA Schemas

• Modeling, Query & OLAP services

• EDW model (master data)

• Authorizations

Operational BI using BW Queries on any loaded data

• Immediate querying on loaded data – no staging to EDW

• Streamlined EDW – releasing EDW from Operational BI scenarios

• Faster time-to-market of BI scenarios

• Incremental build of EDW (from virtual to staged scenarios)

Real time BI (real time replication (SLT) to HANA or to BW )

• Real time master data in EDW (SLT & RDA to InfoObjects)

• Release ERP from operational reporting workload

Free EDW Core from Cholosterol


Open ODS Layer - The Holistic Picture

BW Services for Externally and BW Managed Data

HANA modeler

any schema –

externally managed

Replication

Services

BO Data

Services

DXC

SAP

Extractors

BO Data

Services

Replication

Services

HANA

BW managed schema

HANA Model

BW on HANA provides Services needed to cover the Open ODS Layer

requirements not just for data delivered to BW & modeled with BW (BW

managed) but also to data directly delivered to HANA tables & modeled with the

HANA Modeler (externally managed)


Externally and BW Managed Open ODS Layer Data

Get the Best out of both Worlds

any schema –

externally managed

Replication

Services

BO Data

Services

DXC

SAP

Extractors

BO Data

Services

Replication

Services

HANA

BW managed schema

HANA Model

BW managed & modeled data

Market

(Enterprise) DataWarehouse

Layered Scalable Architecture (++)

Model-driven, Governance, Security, …

Usage

>24,000 active installations

> 16,000 customers

Externally managed data - HANA Modeler

Market

Data Mart

Flexibility, Performance

Native, SQL, tables/views

Usage

HPAs, Accelerators (COPA, Pipeline, …)

Customer-build apps & scenarios

BW-HANA Apps (POS DM, DSiM, …)


A Framework for Source Level Data ?

BW HANA instance HANA

Application

instance EDW

accelerator accelerator

SAP

Application Custom

Model

HANA

sidecar

instance

HANA

sidecar

instance

SAP

Application

ERP2 ERP1


HANA & BI Landscape LSA++ with Open Operational Data Store Layer as Parenthesis

BW HANA instance HANA

Application

instance EDW

accelerator accelerator

SAP

Application Custom

Model

HANA

sidecar

instance

HANA

sidecar

instance

Tight coupling – DTP*

Querying via Virtual/ Transient Provider

& Composite Provider

Soft coupling – remote HANA tables**

(querying via federation)

HANA remote

tables/ views*

HANA remote

tables/ views*

* Planned SP8

** Not available today

DTPs*

tables/ views

SAP

Application

Open Operational Data Store Layer


BW Integration Services

HANA Modeler externally managed Schemas & Data

HANA models in BW

• Consume HANA models

treat them as InfoProvider

BW models in HANA Modeler

• Provide InfoProvider as

views **

HANA schema data into BW

• Transfer data into EDW * BW data into HANA tables

• Transfer data *

(*SP08)

(**Q4 2012)

HANA Modeler

any schema –

externally managed

Replication

Services

BO Data

Services

DXC

SAP

Extractors

BO Data

Services

Replication

Services

HANA

BW managed schema

HANA Model

BW Services for externally HANA Modeler managed schemas & data


LSA++ Open ODS Layer

BW Integration Services for HANA Schemas & Data

HANA Modeler

any schema


HANA Views

EDW Layer

DTP*

Virtual/ Transient

Provider on

HANA Models

Composite / MultiProvider


BW Query

Externally

Managed

BW treats HANA schemas

as Open ODS Provider that

are externally managed

* DTP with SP8


BW Transient and Virtual Provider on HANA Model Different Level of Integration of Field-level Data in BW

HANA

View/Table

Open ODS

Source Model Fields

Transient Model

Transient InfoObjects

derived from Fields

EDW Model

InfoObjects

Transient Model

Transient InfoObjects

Partly assigned

InfoObjects

use Field

properties

may use

InfoObjects

EDW

Core

Provider Virtual

Provider

Transient

Provider

As of SP7 Transient Providers only used in Agile Extensions


EDW Layer

Composite / MultiProvider


BW Query

BW Operational Data Services

Querying on BW Inbound Data (BW 730 SP8)

Open Operational DataStore

DTP

BW enables Operational BI on

BW managed inbound data

filling a HANA DB DSO from

PSA via an Open Hub Service, a

generated HANA view and

Transient/ Virtual Provider(*SP8)

Real time

DTP/RDA

HANA DB DSO*

PSA

Virtual/ Transient

Provider on

HANA Models

InfoObject/

Std DSO/

HybridProvider HANA Views generated

SAP ETL

Data Services

SLT replication


SLT replication: Real Time InfoPackage

BW managed

Open Operational Data Store Layer

BW Operational Data Services Querying on BW Inbound Data New HANA DB DSO (BW 730 SP8)

BW

Data

So

urc

e

PSA

SAP ETL: InfoPackage

DataSource Service -

PSA:

• Consistency services

• Staging services

Externally

managed

BW Virtual/ Transient

Provider on HANA

Models

BW Query, OLAP &

Authorization Services


Open Hub:

• Delivery Services

• Transformation

services

DTPs

HANA DB DSO

HANA Table

Analytic views generated


HA

NA

ta

ble

/

HA

NA

DB

DS

O

BW

Data

So

urc

e

BW Operational Data Services Provisioning, Staging & Query Services - New Operational DSO (*planned 2013)

Operational DSO*

SAP ETL: • DTP on ODQ • Real Time DTP on ODQ


Operational DSO:

• Operational Data Services

• EDW Services

• Consistency Services

• Transformation Services

BW Query, OLAP &

Authorization Services

PSA

SAP ETL: InfoPackage


PSA:

• EDW Services

• Consistency Services

DataSource Service –

Open Hub:

• Delivery Services

• Transformation Services

SLT replication: Real Time InfoPackage

Open

Operational Data

Store BW managed

Layer

RDA

DTP


LSA++ for BW on HANA Value Scenario Incremental build of EDW - from virtual to staged scenarios

Providing Immediate Value of Loaded Data (1st Steps)

BW on HANA is targeting to provide immediate value i.e. Local/

operational BI without obstracting EDW targets allowing to introduce

incrementally an EDW data model (InfoObjects) and the LSA++ when it

shows up value.

Incremental definition

of Data Model &

Architecture


LSA++ Open Operational Data Store Summary

More Flexibility – Immediate Value

The new Open Operational Data Store is the LSA++ Inbound Layer.

The target of the Open ODS Layer is to provide a common framework for BI source level

data.

Open ODS integrates data delivered directly to BW (BW managed) and data delivered to

native HANA models (externally managed)

The Open ODS data are not modeled using BW InfoObjects but inherit the field-level (OLTP)

data model from the source application.

The Open ODS Layer promotes scalable BW Services like Olap services, master data

services , authorization services...

The Open ODS is a hybrid, offering

– Acquisition Layer functionality like request handling

serving as source-layer for the EDW Layer

– Immediate querying** on data arrived via

o SAP extraction (SAPI)

o Data Services for non-SAP sources

o Real time replication via SLT

(*on the same HANA instance) (** HANA DB DSO SP8/ Operational DSO planned for 2013







LSA++ Agile BI



MultiProvider

LSA++ Holistic Architecture Virtual Data Mart Layer - Overview


EDW Layers

Open Operational

DataStore

Agile

DMs/

BW

Work

space

Cen

tra

l B

I

de

pa

rtm

en

tal B

I

LS

A+

+ fo

r BW

on

HA

NA

Virtu

al D

ata

Marts

BW Queries

EPM , BO BI, Apps

Virtual/

Transient

Provider

Analytic

Index

∞ Table/

HANA DB DSO

PSA

HANA Model

Composite Provider

refe

ren

ce

Composite Provider

(** HANA DB DSO SP8/ Operational DSO planned for 2013)

MultiProvider

Operational

DSO



Virtual Data Marts: Virtual Wrapping & Combining Data

LSA++ promotes flexible virtual wrapping of agile / operational data and virtual

combinations (with core EDW context data) using BW on HANA virtualization features

(Virtual Data Marts: Virtual Provider, Transient Provider, Composite Provider)

Virtual Data Marts will increase overall LSA++ framework flexibility and

manageability

EDW context data

Operational

Extension

operational, real

time context data

Flexible Consistent

EDW Core

Agile

Extension

Agile/ ad hoc

context data

LS

A+

+ fo

r BW

on

HA

NA

Virtu

al W

rap

pin

g &

Co

mb

ing

Data

Virtual Wrap Virtual Wrap

Virtual Combination

Query

Vir

tual

Data

Mart

s

Virtual Wrap


LSA++ for BW on HANA Value Scenarios Virtual Data Marts on Open ODS and/ or EDW/ ADM

With LSA++ framework Virtual Data Marts come into focus. Beside the various technical

types of InfoProvider (VirtualProvider, TransientProvider, CompositeProvider,

MultiProvider) we see a wide variety of scenarios addressed by Virtual Data Marts.

Reporting on

Open ODS data

Mixed reporting on

Open ODS &

EDW data

Reporting on

EDW data

Virtual Data Marts flexibility

time-to-market

Persistent EDW consistency – standardized

integration – standardized

robustness, availability – standardized

reproducibility – standardized

latency, freshness - standardized

Persistent Open ODS consistency - source dependent

integration - source dependent

robustness, availability - source dep.

reproducibility - source dependent

latency, freshness – load dependent


LSA++ Virtual Data Mart Layer Structuring Virtual Wrapping & Composition Layer - Naming

Virtual Data Mart wrapping a HANA view versus Virtual Data Marts resulting

from joining (CompositeProvider) other Virtual Data Marts

Please remember the restrictions of Basis & Joined Virtual Data Marts !

Source-level HANA view

Persistent EDW Core InfoProvider

Source-level HANA view

Open ODS

EDW Core

(Virtual) Wrap

Layer

(Virtual )

Composition

Layer

Vir

tua

l D

ata

Ma

rt

La

ye

r

I

W Transient

Provider

Virtual

Provider

Multi

Provider

Composite

Provider







LSA++ Agile BI





BW

Op

en

Se

rvic

es

: D

ata

Mo

de

l,

OL

AP

, A

uth

ori

za

tio

n,

Data

Ag

ing




EDW Propagation

Layer

EDW Transformations



Agile

DMs/

BW

Work

space

Ce

ntr

al B

I- P

rom

ote

To

Pro

du

cti

on

Ag

ile

BI -

ce

ntr

al/

de

pa

rtm

en

tal

LS

A+

+ fo

r BW

on

HA

NA

B

I on S

tream

lined E

DW

,

Opera

tional B

I, Agile

BI

Co

rpo

rate

Me

mo

ry

refe

ren

ce

BW Queries

EPM , BO BI, Apps


LSA++ Agile Data Marts Value Scenarios Central IT-managed Agile Data Marts in Production

Response to short term to business requirements that needs skills

– Ad hoc – single time productive solutions

– Ad hoc – productive prototype with later replacement by promote-to-production

solution

combine & transform data

via APD into Analytic Index

Virtually combine data

via Composite Provider

Immediate querying

via Transient Provider


LSA++ Agile Data Marts Value Scenarios Department (Local)- managed Agile Data Marts (BW Workspaces)

Response to volatile local business requirements

– Local extension of central EDW/ Open ODS core solution

– Local ad hoc solutions

Local independency with controlled access to EDW/ Open ODS core InfoProvider

Resource management

A BW Workspace is a dedicated area in a BW system where new models can be

created based on central BW and local data (flat files). Workspaces intend to bridge

the gap between central governance requirements and local flexibility needs.


Summary

With BW on RDBMS we were limited but the LSA was simple.

With BW on HANA we gain more flexibility and more options covering business

needs – this on the other hand requires a new holistic LSA++ providing a

framework for all different BI-flavours and their persistent data and virtual data

marts

Please remember: we are just at the beginning of a great journey

Thank you

Contact information:

Juergen Haupt

CSA Architect

[email protected]

Documents

LSA++ for BW on HANA - a248.g.akamai.neta248.g.akamai.net/n/248/420835/8bd646f3875ba50bbce... · developments are subject to change and may be changed by SAP at any time for ... Operational