Upload
robert-gleave
View
433
Download
9
Embed Size (px)
Citation preview
BIMODAL IT AND THE JOURNEY TO DATA WAREHOUSE MODERNIZATION
BY
ROB GLEAVE
VP ARCHITECTURE AND STRATEGY
SPORTS AUTHORITY
THE SHIFTING LANDSCAPE OF BUSINESS ANALYTICS
We are all facing this….
THE SHIFTING LANDSCAPE OF BUSINESS ANALYTICS Challenges on all fronts… • "Traditional" IT-centric business intelligence strategies are no longer
sufficient to drive the use of analytics within an organization.
• IT teams are struggling to cope with higher volume and diversity of demand.
• Business is increasingly taking over control, ownership and responsibility for analytics — often underestimating the associated complexity and risks.
• The almost infinite amount of data available offers great opportunity for new insights, but also makes it increasingly difficult for organizations to manage, secure and accurately interpret that data.
THE SHIFTING LANDSCAPE OF BUSINESS ANALYTICS
Without a roadmap to modernization, you will surely fail to keep
up with the demand for data…
What is your plan?
How do you start?
THE SHIFTING LANDSCAPE OF BUSINESS ANALYTICS
First and foremost… Don’t abandon what you have been doing… That is NOT necessary.
JUST
ENTER THE BIMODAL HIGHWAY You can take an alternate route…
BIMODAL IT… WHAT IS IT?
Source: Gartner (April 2015)
!
What is Bimodal IT?
BIMODAL IT… WHAT IS IT?
Mode 1 Mode 2 Reliability, Incremental
Growth Goal Agility, Innovation
Price for Performance Value Revenue, Brand,
Customer Experience Waterfall,
High Ceremony Approach Agile, Low Ceremony
Plan Driven, Approval Based Governance Empirical, Adaptive
Enterprise Suppliers, Long-Term Deals Sourcing Small, New Vendors,
Short-Term Deals Good at Conventional
Process, Projects Talent Good at New and Uncertain Projects
Take the Order, Delight "Customers" Culture Innovate With
"Partners" Long (Months, Years) Cycle Times Short (Days, Weeks)
Think Ninja
Think Samurai
Gartner Presentation, Getting Real About Bimodal, Dave Aron, October 2015
Two Different mindsets and approaches… both essential.
BIMODAL IT… WHAT IS IT?
! Source: Gartner (April 2015)
In a perfect world… we want a balanced mix of both capabilities.
BIMODAL IT… THE CHALLENGE
Most organizations are already executing Mode 1 well, delivering:
Reliability
Efficiency
Safety
Accuracy
The challenge is how to build a Mode 2 capability to deliver
SPEED & AGILITY
BIMODAL IT… THE CHALLENGE
Detractors sometimes discount the idea of Bimodal IT, saying…
“Why not just make every element of IT more agile?”
But that is not realistic…
Bimodal is about accepting deliberate trade-offs
ROADMAP TO MODERNIZATION
The road to modernization will be paved by BIMODAL thinkers
You need a viable plan…
Step #1: Embrace Self Service BI
ROADMAP TO MODERNIZATION
“The users have won…”
- CTO of IBM’s Lean Analytics Division Inventor of IBM Big Insights
Often, IT will resist self service….. WHY? • Business users want it.. • It is often more agile.. • The work of BI is distributed across many hands, who are
all expert in the meaning of the data.. • Self service sandboxes provide real value - they often
become areas of true innovation.. • It helps the ‘underserved’. Most enterprises cannot fund
enough BI to feed the masses.
ROADMAP ROADMAP TO MODERNIZATION
ROADMAP TO MODERNIZATION
Historically with self service, that fear is well founded.. When Sports Authority started the journey to data warehouse modernization, we had:
• “Shadow IT” sandboxes everywhere • Over 1900 MS Access databases on user desktops –
16,800 mdb files (purpose of most were unknown) • One MS Access database supporting 130 users and
containing 50 different tables (many at max size). • Most users queried transactional systems directly to
populate their data marts, impacting performance of those transaction systems dramatically.
ROADMAP TO MODERNIZATION
ROADMAP TO MODERNIZATION
HOST SYSTEM .
ApplicationsApplication
Application
Application
Data Source
Ad-Hoc Report Consumers
Data Sources
Data Source
Application
POS
eCommerce
EnterpriseData Warehouse
Corporate Report Consumers
Scales
Transactional Databases
Scales
Sports Authority - SA Data Landscape
11/1/2015 Page 1
A picture of our world….
Change is difficult, however, IT teams must resist the temptation to “own” everything. The truth is…
• Overly centralized BI teams can't deliver the domain expertise, responsiveness and speed most organizations require.
• While a centralized team does a good job in creating consistency and governance across certain key subject areas, it creates a bottleneck, causing most users to wait too long to get their requirements met.
• The future of BI and analytics is about enabling both a centralized BI function as well as the decentralized analysis occurring within the company.
ROADMAP ROADMAP TO MODERNIZATION
It just takes the right technology…
ROADMAP
Plus, we all know that self-service works…
ROADMAP TO MODERNIZATION
ROADMAP TO MODERNIZATION
Stage 1
What does your technology look like today? Is this your data architecture?
Step #2: Pick your future-state data platform
ROADMAP TO MODERNIZATION
What are kind of environment are we looking for? • Secure • Universally available • Promotes data sharing • Scalable • Reliable • Manageable • Easy to use • Cost-effective
ROADMAP TO MODERNIZATION
Where can you find such an environment? Look to a cloud platform, like Google, Microsoft or Amazon…
ROADMAP TO MODERNIZATION
What is BigQuery? • A fully managed (Saas) data analytics service • ‘Pay for what you use’ model -- very low cost • Familiar SQL Interface • Super-fast! Query against terabytes of data in seconds • Elastic, auto-scaling, up to petabyte-scale databases • Truly ‘Big Data’ – comparable to Hadoop/SQL or Spark • Programmable - APIs, APIs, APIs….
ROADMAP TO MODERNIZATION
How does BigQuery work? • The largest columnar database on earth • Optimized for selection, aggregation • Provides superior data compression • Stores data differently than a traditional RDBMS
ROADMAP TO MODERNIZATION
Step #3: Offer Personal Sandboxes
ROADMAP TO MODERNIZATION
ROADMAP TO MODERNIZATION
Stage 2
Google BigQuery
Google Cloud Storage
Allow users to build personal sandboxes in the cloud environment…
ROADMAP TO MODERNIZATION
Stage 2
At this point, the new sandboxes (Mode 2) sit beside the old EDW (Mode 1)…
Step #4: Experiment with “Citizen” Tools
ROADMAP TO MODERNIZATION
We live in the era of the ‘Citizen’ knowledge worker…
ROADMAP TO MODERNIZATION
Easy-to-use tools are easing business users into functions which have traditionally been handled by IT
• Data blending (e.g. Alteryx, )
• Data quality & MDM (e.g. Alteryx, Reltio, Dell Boomi )
• Data modeling and virtualization (e.g. Looker, Denodo)
• Data visualization (e.g. Tableau, Clikview, PowerBI, Microstrategy 10)
• Lightweight integration and iPaas (e.g. Dell Boomi, Snaplogic)
• Even…. Data Science! (e.g. R, Python Pandas)
ROADMAP TO MODERNIZATION
Generally, these tools are only appropriate for SMEs or people who are currently your ‘super users’.
But, they are extremely powerful and give these business data analysts great freedom to explore, innovate and share valuable data assets.
These tools, extend the reach of IT and actually help eliminate some age-old problems, for example….
ROADMAP TO MODERNIZATION
ROADMAP TO MODERNIZATION
Stage 3
Where do these tools fit into the new data architecture?
ROADMAP TO MODERNIZATION Stage 3
Step #5: Build Core Data Sets in the Cloud
ROADMAP TO MODERNIZATION
ROADMAP TO MODERNIZATION Stage 4
Google BigQuery
Google Cloud Storage
Core Datasets
ROADMAP TO MODERNIZATION Stage 4
Step #6: Migrate to the Data Warehouse of the Future
ROADMAP TO MODERNIZATION
ROADMAP TO MODERNIZATION Stage 5 begins with a full Data Lake, because… • The lake supports direct (Mode 2) discovery against new data sets. • It provides a platform for Big Data – both structured and unstructured. • Its scale-out architecture allows more data to be collected and retained. • It can lower upfront costs, by delaying transformation and modeling until
needed.
Caveats: must define platform, security, data management, etc.
ROADMAP TO MODERNIZATION Notice the full Data Lake and cloud-based Hadoop for Big Data….
Stage 5
Google Cloud Storage
Google BigQuery
Google Cloud DataProc
ROADMAP TO MODERNIZATION Stage 5
ROADMAP TO MODERNIZATION
Join SQL – over user data
set + core data set
BigQuery Sandbox Project
User Sandbox
User Sandbox
Core Subject A
rea
User Sandbox
User Sandbox
Data set secured to an individual user
Core Subject A
rea
User Sandbox
User Sandbox
User Sandbox
User Sandbox
Core Subject A
rea
Core Subject A
rea
Core Subject A
rea
Data set secured to an individual user
Core warehouse data set – read only
Plain SQL – only against user sandbox data set
Sports Authority - SA Data Landscape
11/2/2015 Page 1
Users enjoy a powerful data world consisting of personal sandboxes and core data sets in a new elastic data warehouse.
ROADMAP
In Review, the Steps to Data Warehouse Modernization
1. Adopt a Mode 2 mindset - with a focus on Self Service. 2. Pick a new platform for the future-state data world 3. Offer personal sandboxes for exploration/discovery 4. Introduce and promote “Citizen” data tools 5. Build core data sets in the cloud 6. Migrate completely to the Data Warehouse of the Future
ROADMAP TO MODERNIZATION