During the Chief Data Officer Exchange event in London, Denodo discussed the different ways in which data virtualization can help businesses. In this presentation, our guest speaker Simon Gratton (Deloitte UK), provides information about the emerging approaches to achieving agile data delivery and the cultural issues that stand in our way.
Mapping the pubmed data under different suptopics using NLP.pptx
The 3-Speed Chief Data Officer
1. The 3-Speed Chief Data Officer
Simon James Gratton #ukdata50 #uktech50
2. “The massive decline of PC's…is about society's shift from
creation to consumption… and mass extinction of unnecessary
computer work, which in less than 20 years will be all of it.”
Steve Faktor (IdeaFaktor)
3. A billion tiny robots in the cloud…
…the future of AI in an algorithmic world (CLOUDTECH)
4. 4
Automated Insight needs context….
1. We face a skills shortage
2. AI/Machine Learning can accelerate the end-point
3. Technology needs to bridge from insight-to-action next
TIME
1 MONTH 1 WEEK 1 DAY 1/2 DAY 1 HOUR MINS 1 MIN 10 MS
5. 50%+ of CDOs will be unsuccessful…
…Yet 90% organisations will have one by 2019
6. Today’s data agility is not MVP
A DATA MVP
STARTS HERE
NOT HERE
EMOTIONAL
DESIGN
USABLE
RELIABLE
FUNCTIONAL
7. LINE OF BUSINESS a
LINE OF BUSINESS b
LINE OF BUSINESS … n
PILOTS PROJECTS APPS
ENGAGE, ACCELERATE AND ALIGN
DELIVER (INTEGRATE 1st)
RUN (INGEST 1st)
EXPLOIT (CONSUMPTION 1st)
HOLACRATIC UNITS
Consumption-driven
60%/40% Vertical
OUTCOME FOCUS
SHARED SERVICE
Alignment-driven
40%/60% horizontal
SIMPLIFICATION FOCUS
Balancing Quality and Agility
9. 1. Data needs to be accessible at different speeds and levels of quality
2. Our functions need to be empowered and self-sufficient with tools
3. Our data assets need continuous improvement across our functions
The ‘Value-driven’ CDO
ACCELERATE
DELIVERY
EMPOWER
EXPLOITATION
CONTINUOUSLY
IMPROVE
3.1
1 2 3
13. ‘Fast Data’ means ‘Fast Culture’
We are a collaborative
supply chain of data-
driven value creation
We engage and innovate
side-by-side with our
customers
We positively impact
customer, partner and
employee interactions
We integrate & wrap core
applications for
frictionless experience
We decommission legacy
& repatriate IP with each
delivery
We sit at the front of
change cycle to ensure
integrated outcomes
We drive a single
customer view
harnessing ALL data
We deliver at pace then
continuously improve
once valuable
We drive data innovation
in <90 day agile delivery
cycles
14. The 3-Speed Data Organisation
NEAR-REALTIME SINGLE VIEW
FAST
1-3 MONTHS
Customer Events, Transactions, IoT and Blockchain interaction
3+ MONTHS
The realm of the ‘Data Wrangler’
EXPLOITATION SERVICE LAYER
Wrangler View
Unstructured
Functional view
Semi-structured
Business view
Structured
1ST GEAR 2ND GEAR 3rd GEAR
QUALITY
SLOWMID
<4 WEEKS
UNIFY
CONSOLIDATE AND CLEAN
The Single View
Data Quality Improvements (Graph/JSON)
LOW HIGH
17. ‘In Motion’ = New Approach
CDO FOCUS
TODAY DATA MANAGEMENT
EFFORT
DATA AT REST
EFFORT DATA IN MOTION
EFFORT
CDO FOCUS
NEXT
DATA AT REST
EFFORT DATA IN
MOTION
EFFORT
RELATIVE EFFORT
DATA
MANAGEMENT
EFFORT
AT-DESIGN SCHEMA
STRUCTURE THEN INGEST
PROJECT-DRIVEN
ON-DEMAND SCHEMAGOVERN THE ENTRY-POINT
INGEST THEN STRUCTURE
CUSTOMER INTERACTION-DRIVEN
GOVERN THE END POINT
21. We are in ‘skillset’ transition
2015
INSIGHT COLLABORATORS
80% EXPLOITATION FOCUS
SQL/NOSQLLAKE
R/PYTHON RULES
VISUALISATION ‘WHY’ FOCUS
CUSTOMER-CENTRICITY
FILE GENERATORS
80% INTEGRATION FOCUS
SQLRDBMS/XLS
SAS MODELS
EXCEL/OFFICE ‘HOW’ FOCUS
PRODUCT/LOB-CENTRICITY
2020
‘STRUCTURED’ ‘UNSTRUCTURED’
Wrangler skills are very different to those in use today
Business ‘data workers’ need to be ‘exploiters’ not ‘integrators’
IT ‘data workers’ have traditional skills but limited data science
28. 28
3 Weeks, 3 People, 1 Goal….
BEFORE
29,500 Code Lines
Multiple File Imports
Heavy Preparation. Rerun All
File Storage Overload
Only 10% Code-based removed
Single ‘Virtual’ Feed
Flexible Blending (XML, SQL, File)
Store, Run & Re-run at will
<300
AFTER
5 X Faster,
10 X Cost reduction
DATA ACQUISITION
Multi-User.
1.5 mins per cycle
VIRTUAL WRANGLER PRICING CYCLE
>500
Code-based, Serial Processing, Re-run All 1 User. 5 mins per cycle
COMPLEX FILE IMPORT SAS PROCESSING PRICING CYCLE
DATA ACQUISITION
Direct Simulation Analytics to Actuary