The document discusses SAP HANA, a real-time data platform developed by SAP and Intel over six years. Key points:
- HANA stores all data in memory for real-time analysis without copying or moving data. This allows for queries on billions of records and multi-modal analysis.
- HANA provides up to 1000x faster analysis and processing speed compared to disk-based systems, with 6-1 lower total cost of ownership and operating costs reduced by 30-40%.
- The in-memory platform supports various analysis types including predictive, geospatial, text, and graph/link analysis. It has been adopted by over 6400 customers across various use cases for faster, more advanced
2. HANA Real-Time Data Platform
AN INTEGRATED SUITE OF PROVEN CAPABILITIES FOR DATA MANAGEMENT and ANALYSIS
2
BACKGROUND
§ Result of six year multi-billion dollar co-development effort
between SAP and Intel
§ Released in 2011, there are over 6400 commercial HANA
clients. There is nothing comparable in the market.
HANA ARCHITECTURE
§ Data is entirely in optimized memory
§ Single copy of data – no indexing/aggregates/duplication
§ Bring analysis engines to the data
HANA KEY CHARACTERISTICS
§ Highly compressed enterprise class database for multi-billion
record queries
§ Multi-modal analysis platform for Geospatial, Predictive,
Graph/Link, and Unstructured Text analysis
§ All without copying or moving the data
RESULTS
§ 1000X average increase in analysis and processing speed
§ 6 to 1 SWaP reduction
§ Lower operating costs by 30-40%
4. 4
SAP HANA Platform
Financial
Procurement
Other Sources
Replication
Batch/ELT
Smart Data
Access
Calculation
Engine User Interface
Predictive
Engine
Streaming
Engine
Text Analysis
Graphing
EngineGeospatial
Analysis
CompressionNo Aggregate
Tables
Dynamic Data Tier
Virtual Data
Models
SAP HANA PLATFORM
HANA Studio / 3rd party SQL
Business Objects, SAS,
Clickview, Tableau, Cognos
BI & Analytics Tools
5. 5
Disk based architecture HANA In-Memory Data Platform
SOLVE the enterprise performance problem
A NEW ARCHITECTURAL APPROACH
6. 6
PoC: Text Analysis of News
Articles. Reduced pipeline
time from >24hr to 2hr for
1.5M daily records.
PoC: Multi-modal fusion;
High Performance
Computing replacement
CRADA: Reduced
message processing
time fro 4 hours to
15 seconds
6400+ HANA customers
PERFORMING DATA PROCESSING AND ANALYSIS CRITICAL TO THEIR MISSION
7. Representative SAP HANA Use Cases
• 300 analysts looking for fraud and related
issues
• 500PB Hadoop store of transaction data
• Brought in HANA to introduce automated
anomaly detection
• Analyst productivity rose from 2
confirmed incidents/month to 80
• Social media analytics across 450 sites
• Focused on sentiment analysis (branding)
• Brought in HANA to accelerate analysis,
add geospatial and introduce predictive
analytics
• Queries complete 14,000X faster,
allowing ‘pre-selling’ of trends
• 14000 miles of pipeline they are
required to inspect
• Data stored in Hadoop
• Brought in HANA to accelerate
analysis and to introduce
predictive modeling of trouble
spots (weather, terrain,
materials, etc)
• Analysis time dropped from
63 days to 15 minutes for a 20
mile segment
• Developed Citizen Connect – a
mobile framework for residents
to report on potholes, graffiti,
loitering, etc
• Geospatial and text analysis to
triage the reports
• Prediction of mitigation impacts
• 21% rise in constituent
satisfaction; 89% of citizens
would recommend the app
Analyst Efficiency
Social Media
Analysis
Weather
and
Terrain
Effects
Analysis
Situational
Reporting
and
Analysis
All with a significantly reduced server footprint