why sap hana
DESCRIPTION
Why do we need SAP HANA? What is SAP HANA. High Performace Analytical AppTRANSCRIPT
![Page 1: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/1.jpg)
WHY SAP HANA ?
Ugur CANDANAugust 2011
![Page 2: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/2.jpg)
2
Struggling to Keep Up With Expectations?
![Page 3: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/3.jpg)
3
The Gap Between CPU and GPU
ref: Tesla GPU Computing Brochure
![Page 4: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/4.jpg)
4
The Gap Between CPU and GPU
![Page 5: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/5.jpg)
5
Evolution of Intel PentiumPentium I Pentium II
Pentium III Pentium IV
Chip areabreakdown
Q: What can you observe? Why?ref: Zhenyu Ye / Bart Mesman / Henk Corporaal “GPU Architecture and Programing”
![Page 6: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/6.jpg)
6
Extrapolation of Single Core CPU
If we extrapolate the trend, in a few generations, Pentium will look like:
Of course, we know it did not happen.
Q: What happened instead? Why?
![Page 7: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/7.jpg)
7
Evolution of Multi-core CPUs
Penryn Bloomfield
Gulftown Beckton
Chip areabreakdown
Q: What can you observe? Why?
![Page 8: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/8.jpg)
8
The Brick Wall -- UC Berkeley's View
Power Wall: power expensive, transistors freeMemory Wall: Memory slow, multiplies fastILP Wall: diminishing returns on more ILP HW
Power Wall + Memory Wall + ILP Wall = Brick Wall
David Patterson, "Computer Architecture is Back - The Berkeley View of the Parallel Computing Research Landscape", Stanford EE Computer Systems Colloquium, Jan 2007, link
![Page 9: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/9.jpg)
Big Table
DynamoMapReduceGoogle File System
![Page 10: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/10.jpg)
Fusion ioDrive Octal Capacity 5.12TB 6GB/s Read 4GB/s Write 1,000,000 IOPS
Xpress disk (xpd) 1 TB: 20 GB/s IOPS: 600,000 reads 500,000 writesThe NVIDIA Tesla
c2070 memory 6GB interface 384bit 144 GB/s
8K Camera: 260 km by a fiber optic network 3 GB/sa 20 minute broadcast would require roughly 4 TB of storage
![Page 11: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/11.jpg)
11
RDBMS in MPP Architecture
Each SMB 6.4 GT/s
Useful work of a RDBMS
![Page 12: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/12.jpg)
12
Reporting
“Typical” Business Intelligence Today
Slow
Painful
Expensive
Operational Data Store
Data Warehouse
Indexes
Aggregates
DataBusiness Applications
Copy
ETLCalculation EngineBusiness Intelligence
Query ResultsQuery
Slow
Painful
Expensive
Operational Data Store
Data Warehouse
Indexes
Aggregates
DataBusiness Applications
Copy
ETL
Calculation EngineBusiness Intelligence
Query ResultsQuery
DataMarts
![Page 13: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/13.jpg)
13
In-Memory Computing
Disk is 1Mx slower than direct memory: like a chef doing his shopping on mars
![Page 14: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/14.jpg)
14
In-Memory Computing Costs have Plummeted
Cost of 1 Mb of memory in 2000: ≈$1
Bosphorus bridge:105m / 344ft
![Page 15: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/15.jpg)
15
In-Memory Computing Costs have Plummeted
Cost of 1 Mb of memory today: <1 cent
Child:1.5m
And shrinking….
Dell: Quad 10 Core server 512GB Ram, Hosting Services$5,200/month
![Page 16: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/16.jpg)
16
A Shift of Frontiers in Computer ScienceFreely Adapted from Jim Gray, Turing Award Winner 1998
Tape is Dead Disk is new Tape Main Memory is new
Disk CPU Cache is new
Main Memory
![Page 17: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/17.jpg)
17
SAP ve In-Memory
2010 SAP High-Performance Analytic Appliance
2006 SAP NetWeaver BW Accelerator (BWA)
2004 SAP NetWeaver Enterprise Search (TREX)
1999 SAP Advanced Planner and Optimizer
![Page 18: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/18.jpg)
18
Operations and Analytics Together
Copy
Business Applications
Analytic ApplianceBusiness Intelligence
Add ACID-compliant, row-based, in-memorySingle source of dataFaster, better BI and actionable intelligenceFaster, better applicationsNew application opportunities
Data
![Page 19: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/19.jpg)
SAP HANA
![Page 20: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/20.jpg)
20
In-Memory Computing – The Time is NOW
HW Technology Innovations
64bit address space – 2TB in current servers
100GB/s data throughput
Dramatic decline in price/performance
Multi-Core Architecture (8 x 8core CPU per blade)
Massive parallel scaling with many blades
One blade ~$50.000 = 1 Enterprise Class Server
Row and Column Store
Compression
SAP SW Technology Innovations
![Page 21: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/21.jpg)
21
Fast – Software Optimization for Memory
Conventional databases store records in rows
Storing data in columns enables faster in-memory processing of operations such as aggregates Columnar layout supports sequential memory access A simple aggregate can be processed in one linear scan
A 10 € B 35 $ C 2 € D 40 € E 12 $
A B C D E 10 35 2 40 12 € $ € € $
memory address
organize by row
organize by column
A 10 €
B 35 $
C 2 €
D 40 €
E 12 $
conceptual view
mapping to memory
![Page 22: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/22.jpg)
22
In-Memory Computing – The Time is NOW
HW Technology Innovations
64bit address space – 2TB in current servers
100GB/s data throughput
Dramatic decline in price/performance
Multi-Core Architecture (8 x 8core CPU per blade)
Massive parallel scaling with many blades
One blade ~$50.000 = 1 Enterprise Class Server
Row and Column Store
Compression
Partitioning
No Aggregate Tables
SAP SW Technology Innovations
![Page 23: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/23.jpg)
23
SAP HANA™■ In-Memory software + hardware
(HP, IBM, Fujitsu, Cisco, Dell)■ Data Modeling and Data Management■ Real-time Data Replication■ SAP BusinessObjects Data Services for ETL
capabilities from SAP Business Suite, SAP NetWeaver Business Warehouse (SAP NetWeaver BW), and 3rd Party Systems
Capabilities Enabled■ Analyze information in real-time at
unprecedented speeds on large volumes of non-aggregated data
■ Create flexible analytic models based on real-time and historic business data
■ Foundation for new category of applications (e.g., planning, simulation) to significantly outperform current applications in category
■ Minimize data duplication
SAP In-Memory Appliance (SAP HANA™)
SAP HANA
SQL MDXBICSSQL
SAP BusinessObjects tools Other query tools
SAP BusinessSuite
Other data sources
SAP NetWeaver Business
Warehouse
SAP In-Memory Computing Studio
SAP In-Memory Database
Calculation and Planning Engine
Row & Column Storage
Real-Time Data Replication
SAP Business Objects Data
Services
![Page 24: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/24.jpg)
24
HANA Combines Software and Hardware
In-Memory Computing Engine (Software)
Pre-Installed Systems (Hardware)
+
![Page 25: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/25.jpg)
25
• 2 x 8 core Intel Nehalem EX ( 2 socket system) • 128 GB Main memory • 160 GB PCIe-Flash / SSD for Log volume• 1 TB SAS / SSD for Data volume• 3 x 1 GB n/w or 1 x 10GB n/w (trunk) • Redundant n/w infrastructure
• Uncompressed Data ~ 256 GB to ~500 GB
• Replication Data load 5GB / hr
• 2 x 8 core Intel Nehalem EX ( 2 or 4 sockets system) • 256 GB Main memory • 320 GB PCIe-Flash / SSD for Log volume• 1 TB SAS / SSD for Data volume• 3 x 1 GB n/w or 1 x 10GB n/w (trunk) • Redundant n/w infrastructure
• Uncompressed Data ~ 500 GB to ~1.25TB
• Replication Data load 5GB / hr
• 2 x 8 core Intel Nehalem EX (4 sockets system) • 256 GB Main memory (expandable up to 512 GB)• 320 GB PCIe-Flash / SSD (expandable up to 640 GB)• 1 TB SAS / SSD for Data volume (expandable up to 2 TB)• 3 x 1 GB n/w or 1 x 10GB n/w (trunk) • Redundant n/w infrastructure
• Uncompressed Data ~ 500 GB to ~2.5 TB
• Replication Data load 5GB / hr
HANA Appliance “T-shirt” sizes Specifications & Approximate Data Volumes
S+
S
XS
Starts at S and scales up to M
![Page 26: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/26.jpg)
26
• 4 x 8 core Intel Nehalem EX ( 4 socket system) • 512 GB Main memory • 640 GB PCIe-Flash / SSD • 2 TB SAS / SSD for Data volume• 3 x 1 GB n/w or 1 x 10GB n/w (trunk) • Redundant n/w infrastructure
• Uncompressed Data ~1.25TB to ~2.5 TB
• Replication Data load 5GB - 20 GB/ hr
• 4 x 8 core Intel Nehalem EX ( 8 socket system) • 512 GB Main memory (expandable up to 1 TB)• 640 GB PCIe-Flash / SSD (expandable up to 1.2 TB)• 2 TB SAS / SSD for Data volume (expandable up to 4 TB)• 3 x 1 GB n/w or 1 x 10GB n/w (trunk) • Redundant n/w infrastructure
• Uncompressed Data ~ 1.25TB to ~5TB
• Replication Data load 5GB – 20 GB / hr
• 8 x 8 core Intel Nehalem EX • 1 TB Main memory • 1.2 TB PCIe-Flash / SSD • 4 TB SAS / SSD for Data volume• 3 x 1 GB n/w or 1 x 10GB n/w (trunk) • Redundant n/w infrastructure
• Uncompressed Data ~ 2.5TB to ~5TB
• Replication Data load 5GB – 20 GB / hr
HANA Appliance “T-shirt” sizes Specifications & Approximate Data Volumes
M
M+
L
Starts at M and scales up to L
![Page 27: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/27.jpg)
27
Speed-up Existing Work
![Page 28: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/28.jpg)
28
Discover New Options
![Page 29: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/29.jpg)
29
Customer Testimonials
Consumer and Health Products
Manufacturing
Energy
Communications and Media
Services
Technology
![Page 30: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/30.jpg)
30
The First 35 Years: Innovated with ERP & LOB Apps
ERP + LOBSystems of Record
Data “In” Business AnalyticsSystems of Engagement
BICS Info “Out”
MobilityAccessible Systems
Oracle DB2 SQL Other
Business Applications Performance Bound by Data
ELT or ETL HANAIn Memory Database
ELT or ETLOracleSQL
DB2, etc.HANA
In Memory Database
Three Years Ago: Innovated with AnalyticsLast Year: Innovated with MobilityThis Year: Innovating the DatabaseHANA Accelerates Data, Applications, AnalyticsLong Term: HANA Is the Database
![Page 31: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/31.jpg)
31
The Truth About Enterprise Data Landscapes
Operational
Warehouses
Marts
Dimensional
Semantic
Information
Oracle DB2 SQL Other
BW TeraData Netezza
Mart Mart Mart
OLAP OLAP
IQ
Universe?Queries Ad-Hoc Dashboard
ETL
DATA QUALITY
Applications
Reports
OLAP
Mart Mart Mart
OLAP
Mart
The Value of HANA
HANA
Oracle/DB2/SQL/Other
BW/Netezza/Teradata/IQ
The Future of “The Stack” – HANA Nirvana
HANA
![Page 32: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/32.jpg)
Distribution Center
Retailer ADistribution Center
Supplier
DSD DistributionCenter
Retail Store B
Mfg Plant
Retail Store A
Event Insight
Node
Event Insight
Event Insight
Node
Event Insight
Node
Event InsightNode
Event InsightNode
Event InsightNode
Event InsightNode
SAP BusnessObjects Event Insight
Custom DB
Apps
Apps
Apps
WWW
BW
Apps
![Page 33: Why sap hana](https://reader033.vdocuments.mx/reader033/viewer/2022061120/546d6792af7959ea108b673c/html5/thumbnails/33.jpg)
Distribution Center
Retailer ADistribution Center
Supplier
DSD DistributionCenter
Retail Store B
Mfg Plant
Retail Store A
Event Insight
Node
Event Insight
Node
Event Insight
Node
Event Insight
Node
Event InsightNode
Event InsightNode
Event InsightNode
Custom DB
BW
Event InsightNode
Apps
Apps
Apps
Apps
WWW Event Pattern
Define Event Pattern
Event Pattern Identified
SAP BusnessObjects Event Insight