© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.
Information Management Strategy:
Exploiting Big data and Advanced Analytics
William Dupley
Strategist – HP Cloud
Hewlett-Packard Canada
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 2
Advanced analytics
Traditional BI/MIS/CRM/ data
warehouse SW vendor scope
Big Data: What we are building:
Business
value
Analytics evolution
Reporting
What happened?
Descriptive
Business
intelligence
Why?
Diagnostic
Advanced
analytics
What will happen?
Predictive
Actionable
intelligence
What should I do?
Optimization
Hindsight
Insight
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 3
1. HP Case study: Voice of the Workforce
•
Approach
• Autonomy Explore loaded with VoW comments data
aim discover semantic meaning from the written
comments
•
Key discoveries
• Comment data had not been mined other than key
word search in the past
• Able to determine consistent meaning rather than
interpretation by line managers
• Overall employee base concerned for HP future but
saw the right steps starting
Hypothesis: We will be able to develop more comprehensive corrective plan if we could analyze
VoW results to determine the feeling of the employees
Explore
Insight
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 4
2. HP Case study: Gen 8 Server Call home analysis
Approach
• Aggregate the Gen 8 server call home data
packages. Gen 8 monitors 1600 attributes and
sends encrypted data packages to HP
Method
• Encrypted xml data file sent from server to HP
security gateway
• File is unwrapped and loaded into Hadoop
• Hadoop supports a schema on read feature that
allows to map an SQL structure
• Some relationships and extract and loaded into
Vertica
• SAS is used to Visualize the data in Vertica
Hypothesis: We may be able to reduce failure and outages if we can read the Gen
8 Call home data capture and determine correlations.
C h a t C al l C e n t e r Web & Mobile S o ci al M e di a A u di o V id e o e M ai l N e w s M e di a L o g s C o n fi g S t a t u s
Finance Sales & Marketi ng Service s Supply Chain Relational DW Analytic DW Machine Data Unstruct. Human Data
Quer y Build er Adva nced Analy tics Flexib le Repo rting Interact ive Multidi m. Reporti ng Struct ured Repo rting Score cards Senti ment Analy sis Plann ing & Simul ation Intera ction Optim izatio n & Scori ng Legacy Data Administra tion • Security • Manage ability • Metadat a • Master data manage ment • Quality control Busines s Owned 3rd Party
Business Owned DataOperational Data
• Collaboration • Portal • Dashboard • Mobility • Alerts Data Virtualization Analy tic Appli cation s BI Searc h
Autonomy Hadoop ETL (Informatica) MS Parallel Data warehou se Vertica Hadoop IDOL SAS /R Autonomy Auto nom y Informatica Exc el / Ess base Toa d SQL / Vertica Cust om Business Objects Report Writer Business Business
Data Unstructured Content Data Big Data Enterprise Data Qlik
view Qlik
view
Results
• Relationship have been found between 3
configuration changes that if done in sequence
will produce a service call within 24 hours
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 5
The lines between traditional OLTP, OLAP and Analytics workloads have blurred
The changing nature of BI systems
Sales
Finance
Marketing
Operations
CRM
HR
Sales Targets,
Finance
Targets, etc.
Transactional Data
ETL and Information
Management
Analytics and
Reports/Dashboards
Data extract,
cleanse,
integrate
Homogenized
grain, time
relationships,
dimensionality
Business calculations,
Summarization,
hierarchies
Data
warehouse
Analytics
cubes
Reporting,
analysis &
visualization tools
Extract
(Structured data
source)
Transform
(Build Normalized form)
Load
(Into analytical tool)
Report
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 6
Conceptual end-state architecture
Chat Call Cente r Web & Mobile Social Media Audio Video eMail News Media Logs Config Status
Finance Sales & Marketing Services Supply Chain
Relational DW
Analytic DW
Machine Data
Unstruct. Human Data
Data Mash-ups Interactive Multidim. Reporting Structured Reporting Scorecards Data Access Services Interaction Optimizatio n & Scoring
Legacy Data
BUSINESS ANALYSTS
MANAGEMENT
UNSTRUCTURED
STRUCTURED
Business Owned
3rd Party
Business Owned Data Operational Data
APPLICATIONS/Public Cloud (Aas)
DATA SCIENTISTS
Data Virtualization
Analytic Applications BI SearchData Architecture Enablement
Trusted Data Definition
Structured Information
Management
D
at
abas
e
O
per
at
ion
s
S
ec
ur
it
y
De
fin
it
io
n
Data Virtualization
Data On-boarding
SaaS
Data Access Advanced Analytics Sentiment AnalysisUnstructured Information Management
Big Data
SaaS Integration
Data Services
MDM
Integration Data Store
Data Quality
Industry Data
Elastic
infrastructure
provisioning
Data
access
permission
Data cleansing and profiling
Real Time Response
Advanced Analytics
Foundational Business Intelligence
Insight:
Advanced analytics:
What will happen?
Foresight:
Actionable intelligence:
What should I do?
Hindsight:
Reporting: What happened?
BI: Why?
Analytical Enablement
Load
(Structured/unstructured data/events
Transform
(Build Relationships)
Extract
(Into Presentation/Analytical
Tool)
Illustrate
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 8
Information management architecture example
CO
NS
UM
P
T
IO
N
M
AN
AG
E
M
E
N
T
CO
L
L
E
CT
IO
N
Chat Call Center Web & Mobile Social Media Audio Video eMail News Media Logs ConfigStatus Finance MarketingSales & Services Supply Chain
Relational DW
Analytic DW
Machine/File Data
Unstruct. Human Data
Query Builder Advanced Analytics Flexible Reporting Interactive Multidim. Reporting Structured Reporting Scorecards Sentiment Analysis Planning & Simulation Interaction Optimization & Scoring
Legacy Corporate Data
Administration
• Security
• Manageability
• Metadata
• Master data
management
• Quality control
Business Owned 3rd PartyBusiness Unit
Data
Real Time DB
• Collaboration
• Portal
• Dashboard
• Mobility
• Alerts
Data Virtualization
Analytic Applications BI SearchAutonomy
Hadoop
ETL (Informatica)
Enterprise
Data
warehouse
Vertica
Hadoop
IDOL
SAS/R
Spot
Fire
Tableau
Autonomy
Autonomy ExploreData Virtualization
Excel / EssbaseToad
SQL /
Vertica
Custom
Business Objects
Report Writer
Business unit
Systems
Business unit
Data
Unstructured Data
Big Data
Structured Data
Qlikview
Qlikview
SAP Hana
IDS
Integrated
Data Store
BUSINESS ANALYSTS
MANAGEMENT
UNSTRUCTURED
STRUCTURED
DATA SCIENTISTS
Real Time Response
Advanced Analytics
Foundational Business Intelligence
Foresight:
Actionable intelligence:
What should I do?
Hindsight:
Reporting: What happened?
BI: Why?
Insight:
Advanced analytics:
What will happen?
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. 9
New Technology: Rapid Hadoop Cluster
Development
Sahara provide users
with simple means to
provision Hadoop
clusters
After user fills in all the
parameters, Sahara
deploys the cluster in a
few minutes.
Also Sahara provides
means to scale already
provisioned cluster by
adding/removing worker
nodes on demand.
© Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice.