re-imagine the future in an age of connection and insight · re-imagine the future in an age of...
TRANSCRIPT
10/27/2015 Chart 1
Re-Imagine the Future in an Age of Connection and Insight
Prof. Dr. Christoph MeinelPresident & CEO
Hasso Plattner Institute
University Institute with 10 departments funded by Hasso Plattner Foundation
■ ≈ 150 lecturers, assistant professors and scientific colleagues
■ ≈ 450 bachelor- and master students
■ ≈ 120 PhD students, ≈ 20 external
■ ≈ 240 „HPI Design Thinking Program“ students
USPs: HPI Future SOC Lab | D-School | Research School | openHPI …
Hasso Plattner Institute – Brief Introduction
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 2
openHPI – The MOOC Platform of HPIopenSAP – Powered by openHPI
Prof. Dr. Chr. Meinel Director & CEO
MOOCS – Most important Innovation in E-Learning
■ Massive ■ Open■ Online ■ Courses
Re-Imaginingthe Future
■ Digital transformation refers to the fundamental changes in all branches of human society that are associated and pushed by digital technologies
■ Digital transformation affects
□ our individual life in all its facets as well as
□ all segments of social life
Digital Transformation Changes our Individual and Social Live
DigitalTransformation
Business
Government
Commnication
MedicineTraffic
Science
Art
10/27/2015 Chart 3
Digital revolution, also known as forth industrial revolution is the change from
analog technology,
mechanical technology,
electronical technology to
digital technology
Digital Transformation is Driven by the Digital Revolution
020406080
100120
Change of Technologies
digital
electronicalmechanicalanalog
Re-Imaginingthe Future
■ Digital technologies allow to digitally wrap each entity in the world, whether it is a human or any kind of subject or object
■ This digital wrap, opens a powerful second channel for interactions:
– beside of the traditional physical channel
– a new digital channel is available over the Internet
■ It become possible to interact with any entity remotely via its digital wrap, no direct physical touches are needed. But contrary to physical touches digital remote interactions are possible
– over any distance, and
– almost with speed of light
Each Entity May Become Smart …
Re-Imaginingthe Future
10/27/2015 Chart 4
Internet of Things Allows us to Dream of a Smart World …
Source: http://blogs-images.forbes.com/jacobmorgan/files/2014/05/libelium_smart_world_infographic_big.png
Prof. Dr. Ch. Meinel Director | CEO
Re-Imaginingthe Future
The Internet of Things:Mirroring the Physical World into the Digital World
Source: http://blog.atlasrfidstore.com/wp-content/uploads/2013/07/beecham_research_internet_of_things.jpg
Prof. Dr. Ch. Meinel Director | CEO
Re-Imaginingthe Future
10/27/2015 Chart 5
The Internet of Things:E.g. Smart Home, Smart City, …
Source:http://www.districtoffuture.eu/uploads/imagenes/imagenes_meetinpoint_smart-city_2b637ab6.jpg
Prof. Dr. Ch. Meinel Director | CEO
Re-Imaginingthe Future
The Internet of Things:E.g. Smart Factory / Industry 4.0
Source: http://www.moxa.com/Event/DAC/2013/Factory_Automation_IO/images/Factory_Automation.jpg
Prof. Dr. Ch. Meinel Director | CEO
Re-Imaginingthe Future
10/27/2015 Chart 6
■ Big Data …
■ Cloud …
■ Multicore …
■ In-Memory …
This talk tries to
■ connect these buzzwords and
■ introduces some of our HPI research projects …
Current Developments in Cyberspace are Technically driven by …
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
What Does HPI Do?
10/27/2015 Chart 7
HPI - Top Ranked in Teaching and Research in IT- Systems Engineering
Since 2008, top place among German speaking computer science faculties ...
Prof. Dr. Ch. Meinel Director | CEO
Re-Imaginingthe Future
HPI Concern and Focus:IT-Research with Practical Relevance
Progress in hardware
development
Complex Enterprise Applications
Our focus
Advances indata processing
(Software)
Re-Imaginingthe Future
Prof. Dr. Chr. Meinel Director & CEO
10/27/2015 Chart 8
■ DBMS architecture had not changed over decades
■ New developments make changes possible
□ Hardware trends (CPU/cache/memory)
□ Changed workloads
□ Data characteristics
□ Data amount
■ Meanwhile various database manufacturers have incorporated our ideas and presented new products, e.g. SAP HANA
Recent Developments Allow New Approach to Data Processing
Buffer pool
Query engine
Traditional DBMS Architecture
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
…during recent years, further progress in software for processing data has been achieved
■ Column-oriented data organization (“column-store”)
□ Sequential scans allow best bandwidth utilization betweenCPU cores and memory
□ Independence of tuples within columns allows easy partitioningand therefore parallel processing
■ Lightweight Compression
□ Reducing data amount, while…
□ Increasing processing speed through late materialization
■ and more, e.g., parallel scan / join / aggregation
Recent Progress in Academic Research in Data Processing
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 9
HPI Scientists and Students Have Developedthe In-Memory Database “Sanssouci” SAP HANA
… for enterprise applications
Developed at the chair ofProf. Hasso Plattner
Main idea:
■ Data permanently residesin main memory
■ Main Memory is the primary“persistence”
■ Only one optimization objective: main memory access
■ Cache- optimized algorithms anddata structures
17
Main Memoryat Blade i
Log
SnapshotsPassive Data (History)
Non-VolatileMemory
RecoveryLoggingTime travel
Data aging
Query Execution Metadata TA Manager
Interface Services and Session Management
Distribution Layerat Blade i
Main Store DifferentialStore
Active Data
Merg
eCol
um
n
Col
um
n
Com
bin
ed
Col
um
n
Col
um
n
Col
um
n
Com
bin
ed
Col
um
n
Indexes
Inverted
ObjectData Guide
■ Allow data-centered architecture
■ Serve as the single source for all relevant data
■ Enable real-time analytics of big data
■ Allows informed management decisions on the most current data
■ and many other things …
In-Memory Databases …
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 10
HPI Future SOC Lab: Industry Partners Provide Latest High Computing Systems for Research …
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
HPI Future SOC Lab:Unique Academic IT Infrastructure
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 11
Potentials of In-Memory: Example 1: Personalized Medicine
Personalized Medicine –Multicore und In-Memory Technologies
Personalized medicine
DNAsequencing
Analysis of genomic data
• Quantity: 3.2 million base pairs
• Data size: 1-20 GB
• Quantity:• Known mutations: 80M• Different genes: 20k-25k• Proteins: 50k-300k
• Data size: • Orientation: 5-10 GB• Variants: 10-100 GB
0 10 20 30 40
Supported by HPIToday
Duration (days)
Treatment decisionPatient has cancer Conventional therapy
10/27/2015 Chart 12
HPI In-Memory Genome ProjectChallenge of Gene Analysis
Analysis of gene data
Orientation and Variants Comment analysis inglobal DB
Dependenton CPU performance Storage capacity
Duration Hours – days WeeksHPI Minutes Real time
In-MemoryTechnologie
Multi-Core Partitioning & Compression
HPI In-Memory Genome ProjectReal-time Analysis of Genome Data
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 13
Potentials of In-Memory: Example 2: Cyber Security Analytics - REAMS
■ Cyberattacks exploit (known) vulnerabilities in hardware, OSs and applications
■ The continuous real-time analysis of the various security sensor data makes it possible to detect cyberattacks and to react in real-time time
□ Log files (OS/App), scanning reports, virus firewall warnings, IDS alerts, monitoring logs from different sources, …
□ Post-processing (filtering, compressing…), aggregation, clustering, correlation, visualization
□ Through correlation detection of complex attack scenarios is possible
■ Due to a continuous live analysis immediate responses are possible
Based on In-Memory Technology: Real-Time Security Analysis and Monitoring
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 14
REAMS: Real-Time Event Analysis and Monitoring System
■ Combination of IDS and SIEM
■ Based on SAP HANA Platform
□ fast in processing huge amount of log data
□ sub-second, simple and complex queries
□ HANA analytics capabilities - Predictive Analysis Library (PAL) and R integration
■ Integration of other complex analytics algorithms
HPI-REAMS:HANA-based Architecture
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
Based on In-Memory Technology: HPI REAMSReal-Time Security Analysis and Monitoring System
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 15
Potentials of In-Memory: Example 3: Social Media Analytics
Why do we need to analyze Social Media?
What happens within an internet minute?
■ 100 000 tweets1
■ 300 000 Facebook updates2
■ 80 000 blog posts3
■ people spend nearly 9 hours a day online4
■ 29% of global population are active social media users
□ ca. 60% of North America, 50% of China, 35% of Germany
30
10/27/2015 Chart 16
How Complex is Social Media?
Unstructured Data is everywhere
■ Only personal information like name, birthday, likes and interests are structured
■ Main information is unstructured and buried in
□ Huge amount of posts and interactions
□ Friendship networks
□ Pictures and videos
■ Content-related analysis:
□ Content filtering
□ Opinion detection (opinion shaping)
□ Trend analysis ( “buzz“, “hot topics“)
■ Network-related analysis:
□ Information diffusion analysis / infection tree
□ Communities / blog rings / clusters
□ Rankings
Analysis of Big Data from Social Media by Means of In-Memory Technology
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
10/27/2015 Chart 17
Based on These Principals we have designed a Blog Search Engine: blog-intelligence.com
Prof. Dr. Chr. Meinel Director & CEO
Re-Imaginingthe Future
■ Leverage unknown information on demands in Social Media
■ Become a game changer for communication and sales
■ Identify potential customers and match them across internal and external data
■ Spend more time dedicated for real customers
■ Leverage sales tremendously
Industry Use CaseneXenio Lead Generation
10/27/2015 Chart 18
Relevance based on unstructured marketing material Flyers Presentations Mails …
Industry Use CaseLead Categorization by Machine Learning
Monitoring of several sources
New sources can be added
Internal and external sources
Industry Use CaseReal-time Monitoring of Social Media Sources
10/27/2015 Chart 19
Listen to Sources
Product1
Product2
Product3
Product4
SocialMediaSuite
Knowledgebase for finding discussions
Industry Use CaseneXenio Lead Generation Flow
Industry Use CaseneXenio Lead Inbox
10/27/2015 Chart 20
Future Research
39
■ Deep Understanding – Deep Learning
■ Advanced Forecasting / Mining
■ Artificial Intelligence for Simulated Interactions
Many Thanks of Your Attention
Contact:
Prof. Dr. Christoph MeinelHasso-Plattner-InstitutPotsdam, Germany
0331 5509-222
www.hpi.de/meinel