1310 success stories_and_lessons_learned_implementing_sap_hana_solutions

September 10-13, 2012

Orlando, Florida

Success Stories and Lessons Learned Implementing SAP HANA Solutions

Session 1310

2

About Me

� Jonathan Haun Consulting Manager with Decision First Technologies.

http://www.decisionfirst.com

� Over 12 years implementing BI solutions utilizing Crystal Reports and

BusinessObjects

� Experience with multiple ETL tools and RDMS

� Certified in multiple versions of BusinesObjects Enterprise, BusinessObjects

Data Services and BusinessObjects Reporting tools

� Certified SAP Trainer

� Over one year of experience implementing BOBJ and BW solutions on SAP

HANA

� Manager of the “All Things BOBJ BI blog”. http://bobj.sapbiblog.com

� Twitter Feeds @jdh2n http://twitter.com/jdh2n

3

Learning Points

� What is SAP HANA ?

� Solutions Currently Available based on SAP HANA

� Tips and Tricks implementing solutions on SAP

HANA standalone

� Performance Improvements

� ROI for Solutions Based on SAP HANA

What is SAP HANA ?

5

What is SAP HANA ?

� SAP HANA is a true next generation database. It is a fusion of both software and

hardware that is fully engineered to provide deep and rich access to billions of rows of

data.

� Data on SAP HANA is stored and accessed in RAM which eliminates the traditional

limitations of spinning magnetic disks.

� Data is stored physically close to the CPU allowing faster and more direct access to the

data.

� The HANA software is the first of it’s kind to be built from the ground up to support

multi-core CPU’s and in-memory direct access to data. Memory First, Persistent Storage

Second.

� Data can be stored using either Row or Column based storage providing flexibility in

supporting both Front End Applications and Business Intelligence on a single platform.

� SAP HANA has built in Multi Dimensional Modeling Views that can mimic the capabilities

of OLAP without the need to duplicate data into Cubes. This reduces the TCO of storing

the data while providing the flexibility to perform both MDAS or Operational reporting.

6

Solutions Currently Available based on SAP HANA

7


� SAP HANA Standalone

� Leverage the power of SAP HANA with data from any source.

� Move data into SAP HANA in real-time or in batch using replication or Data Services

4.0

� Highly customizable for any data analysis requirements

� Supports Analytics, MDAS, Forecasting and Operational reporting in a single solution

� Analyze billions of records at amazing speeds using SAP BusinessObjects 4.0

� Similar to the Traditional EIM strategy of building Data Marts

� SAP Netweaver BW 7.3 powered by SAP HANA

� Run BW directly on SAP HANA by replacing your existing RDMS with SAP HANA

� Reduce the overall data footprint of BW with direct integration with SAP HANA

� Easy solution to adopt for existing SAP BW and SAP ECC customers

� Load 3rd Party Data in BW utilizing Data Services


8


� Multiple rapid solutions based on your Industry or Line of Business

� COPA, CRM, Finance, ERP, Cash Forecasting and many others. See more at

sap.com

� Over 20 prebuilt solutions and growing

� Prebuilt solutions that are easy to deploy while reducing development costs

� Ability to manage billions of transactions with sub-second response times


9

� Data Modeling vs Analytic Modeling� Understand the difference before you begin developing solutions on SAP HANA standalone.

Modeling in HANA centers around creating rich MDAS views, forecasting , and complex cross

fact calculations.

� When loading data in batch, model your data into a traditional Star Schema or Fully De-

normalized Table before loading SAP HANA. Joining hundreds of physical normalized tables

will reduce the performance of queries significantly. Also Attribute Views that contain millions

of records should be modeled into the Analytic Foundation tables to increase performance.

� Project Experience:

� 100% Analytic Modeling: Normalized Data loaded into HANA, containing multiple tables and

joins and modeled completely in SAP HANA: 16 sec response times.

� 90% Data Services Data Modeling: With the same data, data modeled into a highly

deformalized table, with Attribute Views containing less then 10,000 records and fewer then 10

total joins produced results in < 1 sec.

� Data Quality must be addressed via business processes and Data Services

� Merging Dimensional Data from multiple sources should be addressed using Data Services

� Reverse Pivots to de-normalize data should be addressed via Data Services

� Hierarchy Parsing should be addressed via Data Services.

Tips and Tricks for SAP HANA standalone

10

� Using More SAP HANA “Analytic Modeling”


11

� Using More Data Services “Data Modeling”


12

� Analytic Modeling Tips� Derived Attributes allow you to make copies of an existing Attribute View. The copies can not

be modified and will always represent the configuration of their master. They are great for

Date and Time based Attributes that are needed to create joins between multiple date

columns found in the Analytic Foundation. When you update the master Attribute all child

attributes will inherit the master’s columns. Similar to creating an alias of a table.

� Joins between an Attribute View and Analytic Foundation can not be defined on Decimal(x,y)

columns.

� Joins on varchar or text columns do not perform as well as joins between Integer columns.

� SAP HANA has a built in Date / Time Dimension (Attribute View). There is no need to use Data

Services to create a calendar date dimension.

� Joining multiple tables in the Analytic Foundation will greatly reduce performance. Better to

use Data Services to combine the tables or Calculation Views to merge two independent

Analytic Views.

� Every Column in an Analytic View must have a unique name. It is best to implement a naming

standard before you begin your end-to-end Analytic View design phase.

� Be cautious making changes to existing Analytic Views (Post Development). Upstream content

will stop working until it is re-activated. In some cases, removing columns, will require re-work

on up-stream content.


13

� Universe Design Tips� Using Analytic Views as the source for your Universe is great when that Universe is used

primarily for Dashboards, Analytics and other reporting needs that contain Group By and

Aggregate functions. All queries executed against an Analytic View must contain a measure.

When you use an Analytic View within a Universe very little Universe design is required as

most of the modeling and metadata were already defined in SAP HANA. It is also a good idea

to leverage Analytic Views as current versions of SAP BusinessObjects Explorer and future

versions of Webi will bind directly to SAP HANA Analytic Views.

� Using SAP HANA Columnar Tables within a Universe (Traditional Universe Design) will provide

more flexibility for a variety of reporting requirements. Best if used for Operational Reporting,

Multi-Fact Data Synchronization and situations where measures are not required.

� Multi-Fact Data Synchronization can be pushed down to SAP HANA by setting the

JOIN_BY_SQL parameter to true within the Universe.

� The SAP HANA ODBC / JDBC drivers are only supported in the Information Design Universe

(IDT.UNX)

� ODBC is easier to setup and maintain on Windows based BOE environments

� JDBC is easier to setup and maintain on UNIX / LINUX environments.


14

� BW powered by HANA� Frequent SAP HANA patching was required during the initial POC. However, the current patch

level (Patch 33) has proven to be more stable and we expect fewer and less frequent patching

as the product matures. SAP has been very diligent in resolving issues but I would recommend

adding time to any project estimate to account for unforeseen issues.

� Hardware vendor selection is important. Watch for vendors recommending supplemental

hardware components. (Network Switches Etc..) Make sure that all supplemental hardware is

compliant with your current environment.

� Frequent SAP HANA database settings were changed during the POC to find the right mix to

support the customers in-house developed BW processes. Loading and processing of complex

BW flows may require increasing parallel treads and memory settings on SAP HANA.

� Some BW modeling optimizations were required to fully support BW on HANA. In most cases

the modeling code was poorly developed (not following best practices) in other cases SAP BW

and or HANA required patches. Note: These issues had little or no impact on the project.

Observations BW powered by SAP HANA

15

Performance Improvements

Baseline Faster Report Speed Faster Report Speed BW Powered by HANAHANA Standalone

Database

Report Response Times: • Reports: • Reports: • Reports: • Reports:

•Avg 90 sec •Avg 40 sec •Avg 32 sec •Avg 6.4 sec •Avg < 1 sec

•CRM 400 sec •CRM 161 sec •CRM 23 sec •CRM 4.6 sec •CRM < 1 sec

•Data Loads 15 hrs •Data Loads 15 hrs •Data Loads 15 hrs •Data Loads 4.6 hrs •Full Loads 3.2 hrs

•Reports are faster then

BWA

•Delta Data Loads are

much faster (3.2x)

BW HANA With BOBJ

4.0 SP4

HANA Stand Alone

With BOBJ 4.0 SP4

•Reports are faster

then BWA and BW

Powered by HANA

•Data Loads are very

fast

BW with Legacy DB2

database. No BOBJ

Legacy DB-BWA With

BOBJ 3.1 SP2

Legacy DB-BWA With

BOBJ 3.1 SP4

•No BWA or BOBJ

Reports

•BWA 2.5x faster

•Early version of BOBJ not

stable

•No Improvement on Data

Loads

•BOBJ 3.1 SP4 very Stable

•Reports are faster due to

Query Striping

Customer’s History & Performance Gains

* Directly Observed during recent POC with customer

16


0

50

100

150

200

250

300

350

400

Base Line BWA BOBJ 3.1 SP2 BWA BOBJ 3.1 SP4 BW HANA HANA Standalone

AVG (Sec)

CRM (Sec)

Query Response Times


17


0

2

4

6

8

10

12

14

16

Base Line BWA BOBJ 3.1 SP2 BWA BOBJ 3.1 SP4 BW HANA HANA Standalone

Load Times(Hrs)

Data Load Times


18

� Reduce Expensive SAN storage by moving data into memory with

compression.

� Better integration of BW with the SAP HANA database appliance

means more operations are managed by the in-memory database

without the need to duplicate data multiple times.

� Multi-Providers on DSOs, in many cases, will work as well (if not

better) then cubes on BWA. Further reducing the data footprint.

� Near Line storage, directly integrated into BW 7.3, can be utilized for

infrequently accessed data. Allows for BW to maintain multiple

storage tiers at different costs levels.

� User will experience faster response times making them more willing

to deep-dive into the data. This will keep them focused on the

Analysis task while increasing their knowledge and productivity.

ROI for Solutions Based on SAP HANA

19


Object Type BW on DB2 Step 1. Cleaning up to move to HANA

Reduction in System Tables

Decommission Obselete Objects

Reduce Layers with HANA

Near line

PSA 5,842 100 100 100 100 100

Legacy DB Overhead 6,141 - - - - -

Change Log 4,174 - - - - -

DSO 1,487 1,487 1,487 1,312 1,201 1,201

System Tables 1,167 1,167 300 300 300 300

Cube 708 708 708 648 591 141

Master Data 408 408 408 408 408 408

Temp 7 7 7 7 7 7

Total (GB) 19,934 3,877 3,010 2,775 2,607 2,157

6 x Comp 3322 646 502 462 434 359

Reduced Data Footprint with BW on SAP HANA

* Directly Observed during recent POC with customer. Your results might very based on the type and makeup of the data.

20


ITEM SAVINGS

SAN (Fast Storage) Eliminated (Very Expensive)

DB2 (Database) Eliminated

Total # of Servers Reduced by 30 -50%

SAP Licenses and Hardware Increased Hardware and Software

Staff Reallocated (DB2 and Storage)

BOBJ No Change

Total Estimated Savings (5 yr) 30-40% reduction in TCO

Overall Cost Savings Estimates


21


Happy and Productive Users

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

User Acceptance

User Acceptance

22

� SAP HANA is a fusion of both software and hardware. It was built from the

ground-up to support in-memory data storage. Legacy vendors are still trying to

develop in-memory solutions based on legacy DB architecture.

� There are three main solution categories for SAP HANA. SAP HANA standalone,

BW powered by HANA and a wide variety of Rapid software solutions.

� For most organizations, SAP HANA solutions will yield significantly faster query

response times.

� There will be an opportunity for most organizations to reduce their TCO for BW

when powered by SAP HANA.

� User Acceptance is hard to measure in dollars, but when users are able to use

BI in seconds and not minutes, they will be more productive and willing to

leverage these tools.

Key Learnings

23

� Data Modeling vs Analytic Modeling in SAP HANA – “All Things

BOBJ BI” http://wp.me/p2868w-4W

� Creating Analytic Models in SAP HANA Studio - Question and

Answer Responses – “All Things BOBJ BI”

http://wp.me/p2868w-54

� SAP BusinessObjects Explorer 4.0, powered by SAP HANA – “All

Things BOBJ BI” http://wp.me/p2868w-2P

� SAP HANA FAQ – “All Things BOBJ BI” http://wp.me/p2868w-N

� https://www.experiencesaphana.com – SAP Hosted Site with

great information on SAP HANA.

� SAP HANA product page - http://www.sap.com/solutions/technology/in-memory-computing-

platform/index.epx

Links and References

Thank you for participating.

Please provide feedback on this session by completing a short survey via the event

mobile application.

SESSION CODE: 1310

Learn more year-round at www.asug.com

Education

1310 success stories_and_lessons_learned_implementing_sap_hana_solutions