4. * Original Question “How should we start a
companywide Big Data adoption? And how
much do we budget? Timeframe – 2-3 years?”
* The Correct Question - “The days of 2-3 year
projects are over. What‟s the fastest, most
incremental way to adopt Big Data which
delivers the biggest bang for the buck?”
*
6. * Big Data in the „real world‟ is mostly about
Engineering.
* Big Data is - about all systems and techniques
to store and process TeraBytes and PetaBytes
of Data
* Big Data might end with Traditional Analytics or
might spill over into full fledged Data Science
efforts.
* Data Science is the science behind leveraging
data
*
7. * Most companies adopt a pilot followed by a Big Bang approach for
company wide adoption. This is the classic Bottom Up approach – you build
the platform, infrastructure, architecture, aggregate data and then open
it up for everyone to use.
* There is absolutely nothing wrong with the above approach but
statistically in the real world that‟s the approach that doesn‟t work unless
you have the worlds top geeks spread all across the company working in
tandem.
* You need to r‟ber three things
* Always take an incremental approach w/ Big Data vis-à-vis an Upfront
Bottom up Big Bang approach
* Identify your strategy, find a few decisions you need to make and work
downwards into Data, Infrastructure, Architecture etc. (Top Down)
* Unless you a Tech Heavy Weight and can pull off a company wide change of
such proportions and also accommodate the costs because your survival
depends on it
* R‟ber you need Early Wins nobody waits 2-3 years for results
*
8. * This is not a Big Data technology presentation.
* There are plenty of those already floating
around
* This deck is about Strategy
*
9. * Today every Data Center sells its services by
calling itself a Cloud (WTH!!! @#!@$#@$)
* 10,000 people DW/BI/Java-Developer Divisions
and basically everyone else on the planet now
calls themself „Data Scientists‟
* Millions of Java/Python/SQL „Application
Developers‟ call themselves Big Data Engineers.
Do you understand the difference between an
„Application Developer‟ vs an „Engineer‟? Do
you?
*
10. * Economics
* Data Economics – The cost of storing say 1PB of Data
* Compute Economics – The cost of processing say 1PB of Data
* And Yes! The ROI
* Value Derived from the Costs of Storing & Processing Data
* And being able to leverage that Strategically
* Most Appliances are …
* Too expensive at scale
* Don‟t scale very well
* e.g. Hadoop has the best Economics & ROI
* You seriously don‟t need very expensive Enterprise Big Data
Software/Hardware/Appliances if your scale involves 4000-1000+ servers to do Big
Data. At that scale you need to seriously contemplate Free-open-source-
software/hardware and take a serious look at
* Economics mentioned above
* And an incremental & elastic approach
p.s. do see my deck on Cloud Computing also in this context
*
11. 1. Historically Businesses has been run based on Anecdotal Evidence
2. DW&BI and currently Big Data Descriptive Analytics give businesses
the „Vision‟
3. Big Data Inferential Analytics give businesses the „Intelligence‟
o The Worlds Front Runners in Virtually Every Industry Segment are the
strongest in Big Data Analytics e.g.
o Capital One, Visa, American Express, PayPal
o Amazon, Walmart, eBay
o Linkedin, Facebook, Square
o Google, Yahoo
o Data is a Strategic Asset just short of being put on a Balance Sheet
*
12. * Mckinsey - 140k-190k analytics positions, and 1.5m data-
savvy managers needed
* Soon a Realization will set in that the existing managers who
make decisions on instinct and experience will mostly not
make the change into Data Driven Management culture and
might have to be let go. Some tough decisions will need to
be made
* Your managers will in high probability internally come up
from the technical ranks who are data savvy. Or externally
from other Technology Majors/Companies who already have
that culture
* Trust me when I say Big Data „Technology‟ is the easy part for
a seasoned technologist and as of today is mostly a no
brainer. The hard part is the Strategic Management Cultural
Shift
*
13. * There are many tactical and operational things you can do with big data. Those should be done in the
second phase after the strategic intent has been achieved and the platform is opened up for everyone
across the company.
* You can also boil the ocean and collect all data, create an elaborate enterprise information
architecture and infrastructure for all eternity. McKinsey taught us not to do that. ;-)
* The answer depends on what‟s strategic to you, don‟t pick prospective projects from cookie cutter lists
floating around for big data adoption in various industries
* Ask – What is our Strategy?
* What decisions do we need to make?
* What data do we need to make those decisions?
* How do we aggregate that data?
* What‟s the minimal setup required to use this data for the above corporate strategy?
* What one or two business functions are the most important for phase #2
* The Plan
* Think incremental,
* Start small,
* Get an early win with the pilot
* Go top down in phase #1
* Go bottom up in phase #2
*