Call Girls From Pari Chowk Greater Noida ❤️8448577510 ⊹Best Escorts Service I...
Externalising Access to BI for Competitive Advantage
1. Externalising Access to BI for
Competitive Advantage
Mike Ferguson
Managing Director
Intelligent Business Strategies
Information Builders Day
London, February 2013
2. About Mike Ferguson
Mike Ferguson is Managing Director of Intelligent
Business Strategies Limited. As an analyst and
consultant he specialises in business
intelligence, data management and enterprise
business integration. With over 30 years of IT
experience, Mike has consulted for dozens of
companies, spoken at events all over the world
and written numerous articles. Formerly he was
a principal and co-founder of Codd and Date
Europe Limited – the inventors of the Relational
Model, a Chief Architect at Teradata on the
Teradata DBMS and European Managing
www.intelligentbusiness.biz
Director of DataBase Associates.
mferguson@intelligentbusiness.biz
Twitter: @mikeferguson1
Tel/Fax (+44)1625 520700
2
3. Topics
Why externalise BI?
Customers, supplier and partner use cases
Planning considerations when making BI available for
external access
The importance of on-demand versus event driven
intelligence
Approaches to making BI available for external access
Performance, security and availability considerations
Conclusions
3
4. The Traditional Data Warehouse And BI Environment Has
Been In Use Within The Enterprise For Many Years
Integration / DQ
P
o
BI
Data
r
Tools web
DW t
Platform
a Reports &
l analytics
Operational Data warehouse
systems Employees
& data marts
Suppliers
Partners
Customers
But what about external user access to BI?
4
5. Why Externalise BI? – Supplier Examples
Retail sales and inventory intelligence to inform suppliers of:
• What products/materials have been ordered
• What products have and have not sold in what stores
• Supplier delivery performance against orders and SLAs
Financial intelligence
• Inform suppliers on outstanding invoices and payment status
Business Benefits
• Supply chain transparency
• Inventory optimisation
• Reduced costs, e.g. lower distribution costs
• Avoids returns, avoids stock-outs and reduces waste
• More timely deliveries
• Suppliers manufacture / ship to meet demand and don’t incur
losses on any returns 5
6. Why Externalise BI? – Partner Examples
Partner marketing, sales and service support
• Clickstream analytics, partner campaign performance reports
• Sales intelligence, e.g. what products are selling where
• Externalise reseller performance intelligence
• Partner field service optimisation
Operations partner support, e.g. Oil and Gas
• Production optimisation
– Production performance reports for partners operating
platforms on behalf on the Oil and Gas company
– Alerts, recommendations to keep operations running smoothly
Business benefits
• Encourage partner growth
• Encourage competition among partners to deliver better service
• Reduce operational cost
• Reduce operational risk 6
7. Why Externalise BI?
– Customer Self-Service Access
Self-service access to billing information and purchases to date
Retail banking example:
• Self-service access to insight on transaction activity and balances
• Provide expenditure analysis and trend reports allowing customers
to better manage their money
Investment banking example
• Customer self-service access to trade accrual reports, cash
forecasting and account positions to facilitate more timely trades
Business Benefits
• Reduced front-office costs through customer self-service
• Builds customer loyalty and improved customer satisfaction
• Provides ‘evidence-based’ cross-sell / up sell opportunities
• Monetisation through value add ‘smart services’ including insights,
predictions, recommendations and alerts to guide customers
7
8. Considerations For Externalising BI
Why do it?
• Define the business case, e.g. cost reduction, better service,
monetisation for new revenue….
What kinds of BI artifacts are you looking to externalise?
• Fixed interactive reports
• Dashboards
• Recommendations
• Alerts
What about data?
• Information services
How will you externalise BI?
• Direct access to BI artifacts
• Access BI via customer facing applications?
8
9. Considerations For Externalising BI – 2
Who will be the external users?
• Customers, suppliers, partners, all of these?
What devices will be used to access BI?
• Personal mobile devices
• Browser access from the desktop / laptops
What BI Applications are needed by which users?
• Customer-facing mobile BI applications
• Supplier BI applications
What is the impact on your existing BI setup?
• Scalability and performance
• Concurrent users
• Security
• Availability
Do you want the users to ACT on the intelligence?
9
10. External BI Can Be Traditional On-Demand Query,
Reporting And Analysis Or Operational BI Or Both
On-Demand Event-Driven
Externally facing Browser Mobile BI
application BI tool/app
Event message, file arrival, pattern, trigger
Application
BI Platform BI Platform
BI service
(query, report, model,
recommendation)
BI service
(query, report, model,
recommendation)
10
11. External BI Can Be Traditional On-Demand Query,
Reporting And Analysis Or Operational Reporting
On-Demand Reporting & Analysis On-Demand Operational Reporting
Externally facing Browser Mobile BI Externally facing Browser Mobile BI
application BI tool/app application BI tool/app
Application Application
BI Platform BI Platform
BI service BI service
(query, report, model,
OLTP
DW (query, report, model,
recommendation) recommendation)
application
11
12. There Are Multiple Approaches To Externalising BI
Integrating BI into externally facing portals
Embedding BI in externally facing processes and
applications
Event-driven push of insights, alerts and recommendations
Mobile BI
It is highly likely that you will need to implement ALL of
these approaches – they are NOT mutually exclusive
Key questions
• Which of these approaches do you need for which users?
• Do you need to use several approaches for the same users?
12
13. Integrating BI Into Externally Facing Apps & Portals - BI Web
Services Have Made “On-Demand” Operational BI Possible
Enterprise Portal
Business Process Management
ESB
Web service API
BI platform
Externally Queries
3. Bind
facing Reports
SOAP Cubes
operational XML results Dashboards
applications Predict. Models
Operational DW &
data Marts
Several BI vendors have had Dynamic discovery and invocation of
this functionality for years e.g. BI web services – this is an industry
Information Builders Service Registry standard approach to BI integration
13
14. Externalising BI – Service Enabled BI Platforms Make
It Possible To Integrate BI Into Portal Technology
Customer /
Supplier /
Partner
Portal
BI Tools Platform & Services
DW/
Marts
Integrate the BI Services directly into an enterprise portal
Many vendors provide pre-built widgets for popular portal technologies
14
15. Business Optimisation Requirement: BI Needs To Be
Integrated Into Operational Business Processes
Integrated Intelligent Business Operations
Shipping/distribution
Finance/accounting
Human resources
Inventory control
Service/support
Procurement
Operations
Marketing
Sales
Customers Partners &
suppliers
Integrated On-Demand Business Intelligence
Customer Supply
Operations
relationship chain
management
management management
Employees
15
16. On-Demand Partner BI - Integrate BI Into Process Activities To
Guide Operational Decision Making During Process Execution
Order Entry, Fulfilment and Tracking Process
Which process activities are
performed
• automatically by software How can BI be leveraged
• manually by people, to help improve business
• by people using operational apps BI on-demand performance in specific
• by people who are mobile ? process activities?
16
17. On-Demand Recommendations Can Be Used To Cross Sell /
Up-Sell Or Simply To Improve Loyalty And Satisfaction
operational historical
feeds data data
DW
data virtualisation
(information service)
Externally automatic analysis
Facing (predictive /
scoring models)
Operational
applications
real-time
decision (rules)
on-demand engine
recommendations
Recommendation Service
17
18. “Smart” Applications - Leveraging Deployed Analytic
Models In DBMSs And Operational Applications
Many BI vendors have
supported this capability Customer, Business
for several years Partner, analyst
Supplier
Data mining operational application BI tool or
analyst SQL SQL application
Data Mining Tool Operational
Transformed Data
data
Assimilated
Extracted Information
Information
Data
Warehouse
Selected
Data UDF data warehouse
Select Transform Mine Assimilate
PMML
mining e.g. scoring DBMS
model model
embed
Historical model operational
data application
e.g. generated code model
classification model
18
19. Event Processing – Automated Analysis And Automated
Decisions Can PUSH Alerts And BI To Mobile Devices
operational historical
feeds data data
DW operational
event driven low data streams
latency data
External mobile Data Integration
capture
users
In-memory data
automatic analysis
(predictive / Balancing Act
scoring models)
automated You can’t afford to
alerts & BI annoy good
pushed to customers however
Auto real-time
the device decision (rules)
recommendations can
engine improve loyalty and
satisfaction
event driven ‘decision service’
19
20. Right Time Business Optimisation Means Operations
Must Act BEFORE Data Reaches A Data Warehouse
We are moving towards an event-driven enterprise where every
transaction and message is an event
Events need to be captured and analysed and acted upon in time to
keep the business optimised and to build trust with external users
Requires monitoring the pulse of business as it happens
20
21. Classic Data Warehousing Vs Event Processing
- The Architecture Has To Change
Classic DW Data quality Store data Analyse
& integration (human)
Event Event Data Analyse Act Store
processing Detection quality & (automated) (automated data
integration or human)
Data Quality of event data is key to ensuring that automated decisions are accurate
21
22. What is Event Processing? – Find & Analyse Event
Correlations In An Un-Sequenced Event Storm
capture
Event Sources
CEP Runtime
Monitor events from multiple Evaluations
sources (feeds, databases, Correlations
sensor networks, RFID
scanners,..) simultaneously
looking for event patterns
?
external event Internal event
sources Automated Action
sources
“Potential Fraud!”
“initiate action
22
23. The Enterprise Data Quality Firewall – On-Demand,
Event-Driven And Timer-Driven Processing
Access
applications
via the portal
or direct use
Enterprise Portal
of UI
Operational apps DW BI
Operations ERP Tools MDM system
D
BAM R
CRM Supply I
Tools C
Chain U
D
external Enterprise Service Bus
msgs (message queuing, routing, translation)
Service API
Import/ Vocabulary definition,
export Enterprise
DQ Server metadata discovery &
DQ mapping
Process Repository DQ Assessment
Management (rules) DQ Rule development
Sources DQ Admin
23
24. Rules ALSO Have An Important Role In Operational BI
operational historical
feeds data data
On-Demand Recommendation
DW
data virtualisation
(information service)
Event-Driven Automated Action
automatic analysis
Operational (predictive /
scoring models)
applications operational historical
feeds data data
real-time DW operational
decision (rules)
on-demand engine data streams
recommendations Data Integration
Recommendation Service
In-memory data
event driven low
automatic analysis latency data capture
Automated (predictive / scoring
user alerts models)
real-time
External users decision (rules)
engine
event driven ‘decision service’
24
25. Achieving Consistent Customer Treatment Via On-Demand
Access To Common Recommendation And BI Services
This architecture applies to any externally facing customer, supplier or partner applications
Front-Office Operations
E-commerce Customer Sales Force Customer facing Marketing
application service app automation app outlet applications application
Enterprise Service Bus
Common Common customer Improve marketing, sales
customer recommendation and service via on-
BI services services demand access to BI and
Clean, recommendation services
DW integrated,
BI System
relevant to each customer
historical data
Data Management Platform (capture, clean, integrate, load)
Operational services
systems RDBMS Files Feeds spreadsheets
25
27. Case Study: Large Bank Using On-Demand Decision Services
In A SOA For Customer Level Risk Management
Customer portal
Call Centres Branch Systems
Front-office Front-office
application application
Enterprise Customer Risk
Management DW
Decision (all customers, all risk
Business
products with scoring
Process Decision
service models in the DBMS
Server Decision
service invoked by a scoring
service DW service)
Service Bus
Back-office Back-office
application Back-office application
Mortgage System application Credit Card System
Loans System
27
28. The Demand For Smart Operations And Business Optimisation
Is Bringing BI To The Centre Of The Enterprise In A SOA
Enterprise Portal with Collaborative Workspaces Office
PM applications applications
Business process management Search
(Scorecards,
operational operational dashboards,
App’n services App’n services budgeting,
planning)
Leverage BI to drive and
guide business operations
Common BI & Recommendation Services
Service bus
Common BI Platform
CEP agents & registry
& rules engine trusted,
predictive analytics
DW Integrated,
External historical data
events
Data Management Platform (capture, clean, integrate, load)
Operational Web services
systems RDBMS Files Feeds spreadsheets
28
30. Mobile Devices – US And Western Europe Mobile Device
Revenue Increasingly Driven By Applications And Data
Source: Mobile Industry Predictions Survey, Chetan Sharma Consulting, 2012
30
31. Development On Mobile Devices
– Apple, Android And HTML 5
Source: Appcelerator/IDC Mobile Developer Report 2012
31
32. External Mobile BI – Targeted BI Content Must
Support ANY Device
iPhone Android Blackberry Symbian Windows
How many mobile Mobile BI Server What is the cost of
operating systems supporting any and all
will you support? devices?
BI Platform
Do you need Do you have the skills in-
browser-based or place to support customer
Hybrid mobile BI for problems on every device?
external users?
DW
What happens when a new
device becomes available?
32
33. What Is HTML 5?
Scalable Vector Graphics – Draw on your web page and make the drawing interactive
Canvas API – add on top of your graphics and control interactivity
Article Mark up parts of the content that is independent, e.g. blog post, article etc.
Aside Used to mark up relevant additional information, like a sidebar
Audio Used for natively including audio in a web page
Header Used for headers in its context. Note: not just for the header of a page, but also
for each header part in section, article and similar
Nav Used for marking up main navigation.
Section Mark up a generic document section. Easily confused with article, and on top of
that you nest either of them, in any order, with the other
Footer The counter-part to header; use for any footer section per context
Video Used for natively including video in a web page – lots of interesting work is
coming along here in terms of web browser support
App Cache and Database – Store data on the device for off-line connectivity
Web workers – execute Javascript in the background rather than kill the browser (these
Javascript functions can’t access the screen but can do server requests
33
34. What Should Be In A Mobile BI Strategy?
Which devices do you need to support?
What mobile devices operating systems need to be supported?
• E.g. Apple iOS, and Android
What is your corporate policy on mobile devices?
• Device standardisation Vs device freedom?
Who is your target user audience for mobile BI?
• e.g. Executives, operational users, external users
What kind of BI content do they need on what kind of device?
What do you need to do to make existing BI content available on mobile
devices?
What do users need to do once they have received the BI content?
How will they act on the intelligence from a mobile device?
Skill sets, help desk support
Catering for concurrent user scalability 34
35. Mobile BI - Ease Of Use Impacts The Number Of
People Using BI
Mobile BI tools
should demand
minimal user
input via the
mobile device
Source: Ease of Use and Interface Appeal in Business Intelligence Tools
Cindi Howson, BI Scorecard
35
36. Mobile BI Requirements
- Mobile Devices And BI User Functions
Mobile BI Authoring? BI / Information Share/ Present Act on BI from
Device Consumption? Collaborate BI to Mobile Device
over BI others? by Initiating
Content? Transactions?
No Yes Yes No Yes
(mainly BI (Email,
‘Nuggets’) SMS)
Alerts, KPIs
Recommendations
Yes Yes Yes Yes Yes
(re-authoring -
summarise &
re-publish)
Yes Yes Yes Yes Yes
36
37. Mobile BI Strategy - Key Considerations
Mobile devices and operating systems supported
Product architecture, e.g. connect to 1 or multiple BI platforms?
Content and data sources accessible
Authoring mobile BI content
Consuming mobile BI content
• Who are your mobile users and what should they be able to do?
Performance
Security
• Mobile devices get stolen, lost…
Integration with other applications and services
• Ability to share and collaborate over BI content from the device
• Ability to act on BI from within the mobile BI app on the device
Packaging and pricing
37
38. Mobile Interactivity – What Is It That You Want
External Users To Be Able To Do?
Prompted reports
Refresh, e.g. to recalculate metrics?
Sort
Drill
Filter
Change chart display (FLASH problem on iPad, e.g. Heat Maps)
Auto rotate
Maps
Use of barcode as input?
Magnifying glass
Which gestures are good when you need to do these things?
Create additional metrics?
38
39. The Ability To Act Is Already Embedded Inside Many
Mobile Apps Already - E.g. In-App Purchase
Act inside the application
Evidence indicates this is well worth doing
Source: www.flurry.com/blog
Revenue from in-app purchase is exceeding
advertising revenue
39
40. You Get The Mobile BI On-Demand
– OK…So Now What? - People Need To Act On BI
E.g. Request a Demand versus
Inventory report for a scanned product Get a Mobile BI report
Invoke a
transaction
e.g. Information Builders MAINTAIN Re-Order Inventory
Transaction
40
41. Decision Management And Mobile BI
We need to embedded recommendations and actions in
mobile BI to help people act quickly
You have to decide
• What actions should be taken from within Mobile BI applications
• What alerts and recommendations to push to guide people into
taking action
Embedded actions could include being able to
• Purchase, calendar event, share, track, email....
• Mobile BI Apps that tempt customers
• Mobile BI Apps that guide employees / partners / suppliers
• Mobile BI Apps that have embedded collaboration
– several Mobile BI Apps already have email embedded
Actions need to be logged for auditing, compliance and
management purposes 41
42. External User Access To BI Has Significant
Implications On BI Systems
Can your BI
system(s) scale
to handle
Enterprise Portal
external users?
Business Process Management
ESB
Web service API
BI platform
3. Bind Queries
Operational Reports
SOAP
application XML results Cubes
Mining Models
Operational DW &
Concurrent user
data Marts
scalability,
Mixed workload,
Performance
43
43. New Advances In Analytical DBMS Technology Also
Help Address The Scalability Issue
Columnar data In-memory data
Database
Server
In-Database Analytics
Data Predictive
Data Prep
Mining Analytics
Multi-temperature R Scientific
Spatial
data Analytics Analytics
Solid State Disk
44
44. Query Performance
- Pushdown Optimisation And In-Database Analytics
General trend to running analytical function much closer to
the data to improve performance
• Often referred to as in-database analytics
Objective is to achieve ‘pushdown optimisation’ to exploit
powerful analytics in the DBMS rather than take big data out
and do this outside the DBMS, e.g. as UDFs
Examples of in-database analytics include
• Predictive and statistical models running inside the DBMS
MapReduce external table functions running in the DBMS server
• Analytical functions written in other languages and running in the
DBMS
In-Database Analytics
Data Data Predictive
Prep Mining Analytics
e.g. IBM Netezza
R Scientific
Spatial
Analytics Analytics
Picture source: IBM Netezza
45
45. Predictive Analytics In The DBMS Is Key To Scalability Of
Concurrent User Requests For On-Demand Recommendations
Business process management
operational operational
application application
SQL or web service requests
DW
Works well IF all the
predictive analytics
Predictive/statistical data needed is already
in the DBMS
models in the DBMS
Rules engine
DBMS server
46
46. On-Demand Recommendations May Need Data Integration
And In Memory Data – Making Use Of Information Services
Recommendation Service
real-time
on-demand decision (rules)
recommendations engine
automatic analysis
Operational (predictive /
applications scoring models)
In-memory data
In-memory data and pre-integrated data caching in
are important considerations for performance
external user access DW
feeds operational historical
data data
47
47. In Memory Technologies – Many Products Are Now
Exploiting In-Memory Capability
BI Platforms and tools to facilitate rapid interactive reporting
and analysis
Analytics for rapid predictions and recommendations
Analytical relational and multi-dimensional in-memory
databases for handling much larger numbers of concurrent
users accessing externally facing BI tools and applications
that invoke pre-defined queries on-demand
48
48. Conclusions
Business drivers will dictate external user access to BI
External users want ‘value add’ BI and information services
that help them
• Insights to increase loyalty and satisfaction as well as cross-sell
Segment you external users and define what BI and
information services you want to target at these segments
• Interactive reports, dashboards, recommendations, alerts….
Consider multiple delivery models for BI services
• E.g. Mobile BI, embedded in self-service apps, via a portal…
Upgrade and test existing BI systems to meet scalability and
availability requirements
Monetise where there is real value add and convenience
Enable external users to act on the intelligence they receive
49