Harness the power of data

Now is the time to become an analytics-driven organization.

Discover how.
INTRODUCTION

Imagine this...
…You are reviewing the weekly spreadsheet that’s just been delivered
to your email. You spot an anomaly in the financial data that you don’t
understand—despite the pivot table provided in the report that allows
you to drill down to at least some level of detail.

You ask your operations analyst what’s going on. To which your analyst
responds, “I’m not sure. Let me find out.”

The next day, the analyst tells you that the reason for the anomaly is
the fact that productivity was way down at the manufacturing plant.

“That doesn’t make sense,” you say. “Can you ask Human Resources if sick days
are impacting the productivity numbers? Or could it be that there was an
issue with the time-capture application at the plant?”

“It will take a week to get that data and merge it with the financial data,”
your analyst says.

“Can’t you just send me a dump of the data from the ERP and time
application, and I’ll work with it myself?”

The analyst responds,

“I don’t have access to the data, and it will take a few

days to submit the right tickets to get access to it.”

2
DATA ARCHITECTURE BEHAVIOR

How does data architecture behave?

Sound familiar? The key fact is that every organization has a Just as nerve endings extend throughout the
growing data problem. First reactions often lead human body, sending sensory signals to be
Your organization companies to believe that these issues are related processed by our brains, data architectures move
is not alone. to business intelligence processes and tools. information across modern networks. They have
However, the real hurdle is the inability to simply the ability to receive, process, and store data from
assess the data that enterprises already have. anywhere, inside and outside the enterprise. The
With artificial intelligence and machine learning signals are processed in real time and acted upon
capabilities coming to fruition, now is the time by machine learning algorithms. These capabili-
to harness the power of information to become ties are applicable across the board—not just to
a data-driven organization. specialized data scenarios.

3
DYSFUNCTIONAL DATA

Harnessing data means Big data problems are about more than just
removing the dysfunctions volume. Most organizations will encounter some
or all of these common data dysfunctions.

Decentralized data Dark data

Siloed data exists across many applications, storage system You have data, but not all of it is loaded into your data warehouse.
databases, and data warehouses. Storing data requires its own Data warehouse capacity is limited and unstructured data is difficult.
access roles, rules, and processes. Your data model does not have the right fields.

How do you derive meaning from all of your data when you How do you get insights from all your data when all your data
cannot access all of your data? is not being analyzed?

Diverse data Delayed data

New data types are being created: Internet of Things sensors, Batch data loading is still pervasive. Nightly batch jobs involve
images, videos, social media, text, and documents. Legacy systems extracting data from one system, transforming it, and loading it into
have issues handling these new data types. Data can now be another system. This transferred data can be analyzed or used for
semi-structured or unstructured. Diverse data storage and analysis another purpose. The transfer of data creates delays for data availability.
differs from that of traditional relational or transactional data.
How can you make a decision based on up-to-the-minute data
How do you store and analyze new data types with analytics if the data is not available until a later time?
tools that are not optimized for new data types?

4
DYSFUNCTIONAL DATA

Duplicate data Deleted data

Data silos may store duplicate data. Different departments store Interesting data may be deleted. Older data that may seem to
and use customer data for different purposes. Each department provide little use today could be deleted. However, teams may
may use unique customer data. Some data may be common across need to analyze deleted data.
these departments.
How can you take advantage of data if it has been intentionally
How do you determine which system to get the duplicated or unintentionally deleted?
data from?

Deserted data Dirty data

Data may be stored for a new project. Data may be deserted To confirm data is as clean as possible before being stored
when the project is cancelled, priorities shift, and people change and analyzed, forms, rules, and other validations are necessary.
jobs. This data can still have purpose, meaning, and be useful. Interesting data, such as unstructured or object-based data,
However, people may not be aware of what the data is or its value. may not be so clean.

How do you incorporate valuable deserted data into How do you explore dirty data to understand what value it
your analytics? can provide?

Conclusion
If these points are familiar to you, it may be time to revisit your approach to data
and analytics, and how your architecture enables it. You can now deploy fit-for-purpose
analytics solutions (storage, processing, querying, analysis, presentation) to meet
existing and future business and IT challenges.

5
DATA LAKE ARCHITECTURES

Modern analytics Access to any and all data With data lake architectures, you can run any type
of analytics—from simple SQL queries to com-
platforms enable critical Data-driven decisions necessitate access to all
plex data warehouse, big data, machine learning,
business insights your data—areas, systems, applications, and data
sources—by those who need it, when they want it.
and predictive analytics. Data lake architectures
decouple compute from storage so that you can
scale either as needed. They can also implement
Once you’re ready to tackle your storage tiering to optimize data availability and
Store it all in one place—an
big data problem, what can you organization-wide data lake. storage costs.
reasonably expect to accomplish The advantages to data lake architectures:
with a modern analytics platform? • Store any data, in any format
• Durable, highly available
• Extends to petabyte and exabyte scale
• Secure, compliant, and auditable
• Handles regulatory and privacy
requirements

6
CASE STUDY/FORTNITE

Keeping sophisticated How revenue is achieved Epic Games collects billions of records on a
daily basis, tracking virtually everything hap-
gamers engaged Fortnite is free to play, with revenue coming en-
pening within the game: how players interact,
tirely from in-game microtransactions—meaning
how often they use certain weapons, and even
Fortnite is an online game developed and run its revenue depends on continuously capturing
the strategies they use to navigate the game
by Epic Games. In the world of Fortnite, players the attention of gamers through new content and
universe. It’s a massive amount of information,
can cooperate on missions or attempt to be the innovation.
and it’s all stored in their data lake for analysis.
last person standing in the game’s Battle Royale When the company introduces new elements
How Epic Games harnesses the power
mode. The game has become a phenomenon, and updates, it can use analytics to almost
of data
with more than 250 million players around the instantly learn how the Fortnite community is
world supporting over 10 million concurrent play- Epic Games needs an up-to-the-minute under- engaging, and then respond accordingly.
ers during in-game events. standing of gamer satisfaction in order to cre-
ate an experience that keeps players engaged. Watch the presentation >

7
WHY A MODERN ANALYTICS PLATFORM?

Use purpose-built analytics

tools on the same data
Traditional data warehousing vs. modern data warehousing
Why use an analytics platform? Simply stated,
one size does not fit all data or use cases, and Traditional data warehousing and visualization tools may fit basic reporting,
dashboards, and ad hoc queries on structured data that resides in data warehous-
using separate tools to analyze disconnected data es and data marts. Some use cases require access to data that is not in your data
is complex. A modern analytics platform can handle warehouse and analyzed in context with the normalized data in the data ware-
diverse data types, wide-ranging use cases, and house. A modern data warehouse can extend data warehouse queries to include
data residing in the data warehouse, the data lake, and operational databases.
multiple analytics approaches on an organization- This happens without any delay of data transformation or movement, so you
wide data lake. always get those essential timely insights.

Can other analytics tools access the same data lake?

What do modern analytics tools need?
Other tools like big data analytics on unstructured data, real-time data stream
They need to access and process any data from the same source—your data lake. analytics, and predictive analytics require purpose-built analytics tools that can
access the same data in the data lake. This allows everyone across the organi-
How does it access data? zation, from business users to data scientists and everyone in-between, to have
Many analytics tools can access and process data directly from a data lake. confidence in both the data and their analytics results.
Data in your data lake can also be queried from a modern data warehouse to
take advantage of a structured and normalized data model with a high-perfor-
mance query engine.

8
CASE STUDY/EQUINOX FITNESS

Keeping customers constantly motivated

Equinox Fitness—a company with integrated lux-

ury and lifestyle offerings centered on movement,
nutrition, and regeneration—replaced its traditional
on-premises data warehouse with a cloud data lake
analytics platform.

The system collected data for their cloud data lake and data warehouse from:

• Exercise equipment (gamified cycling machines, cardio machines, etc.)

• Clickstreams
• Cycling logs
• Service-enhancing software

How Equinox Fitness builds on purpose-built analytics

Purpose-built analytics tools perform simple and complex queries against
both structured and semi-structured data. Developers, business analysts, and
data scientists throughout the company are empowered with the insights
they need. Customer experience, engagement, and overall satisfaction have
improved as a result.

Watch the video >

9
REAL-TIME DATA AVAILABILITY

Real-time data availability for analytics

The speed at which decisions can be made is dictated Key facts

by the availability of the data to be analyzed; and it • Ingesting data into your data lake or data warehouse in periodic batches,
typically each day, delays time-to-insight.
needs to be extracted, transformed, and loaded before
it can be analyzed. Therefore, we should expect • You should never have to wait for the daily batch to be completed to
understand where your business stands.
modern analytics systems to ingest and analyze data
in near real time and be responsive to changes from • Streaming technologies allow data to be ingested at a massive scale, in
real time, so you can perform analytics almost instantaneously.
upstream data sources.

10
CASE STUDY/NETFLIX

Data and analytics on a massively Real-time response across TBs of daily data
entertaining scale Netflix must constantly monitor and optimize its network in order to con-
tinuously improve customer experiences, increase efficiency, and keep costs
under control. By using Amazon Kinesis Data Streams to ingest, augment,
Netflix—the world’s leading internet television and analyze multiple terabytes of daily data, Netflix can quickly identify and
network—uses data and analytics services from act on opportunities to maximize application uptime.

AWS to help keep its 100 million+ members reliably

Creating a better streaming experience
and happily streaming a collective 125 million hours
Netflix’s Amazon Kinesis Data Streams-based solution has proven to be
of content each day. highly powerful and scalable. It processes billions of traffic flows each day,
generates updated analytics within seconds, and helps ensure high availabili-
ty and great experiences for Netflix customers.

Read the full story >

11
MODERN ANALYTICS

Interactive insights where and how you want them

By considering the user experience when interacting with data,

Business Intelligence (BI) tools have made things better. No
longer do users have to suffer frustrating analytic experiences,
delays, or complex searching for the data they need.

However, the number of tools provided to organizations requires users to be experts in each
one. With infrequent use of tools, users experience frustration and costs can become exorbitant.

A modern analytics platform should empower users by providing interactive reporting,

dashboards, as well as easy and intuitive visualizations. The platform should also have flexible
licensing with pay-as-you-go terms and come with the freedom to explore the data without
incurring large IT investments in infrastructure and licensing.

12
USE CASE/NFL

Keeping data visualizations

in front of decision makers

The National Football League is a professional

American football league consisting of 32 teams,
divided equally between the National Football
Conference and the American Football Conference.

Modern business intelligence (BI) tools allow the NFL to quickly build fast, in-
teractive dashboards that seamlessly integrate with its Next Gen Stats applica-
tions. The NFL extends these secure, customized, and easy-to-use dashboards
to each Club without having to provision or manage infrastructure, all while
only paying for actual, pay-per-session usage.

Read the NFL story >

13
ARTIFICIAL INTELLIGENCE

Intelligence embedded in
the business
Artificial intelligence and machine learning are
leading the way. Advances in machine learning
frameworks, coupled with the use of specialized
servers utilizing graphics processing units (GPUs),
are enabling all kinds of new capabilities, like
autonomous driving.

Where does data come in?

In order to train machine learning models, vast
amounts of data are required. Using a data
lake can make it easy to store data and make it
available for AI and ML. Organizations are already
starting to build these capabilities to drive new
outcomes not previously possible, such as being
able to better predict health outcomes based on
retinal imaging.

How do you integrate algorithms with

business processes?
Getting the insight or data science model created
is the easy part—getting it integrated into your
insurance policy engine or retail platform is the
hard work. These systems do not typically have
the ability to integrate outside data sources or
APIs. This is a great opportunity to consider moving
these systems to the cloud to take advantage
of all the services available to help modernize or
re-architect them.

14
CASE STUDY/ZAPPOS

AWS helps Zappos tie personalization

and data together
Zappos began 20 years ago as a small online shoe
retailer. Since then, it has grown to sell clothing, hand-
bags, and accessories—while becoming renowned for
its exceptional customer service. The company has
been a subsidiary of Amazon since 2009.

Searching for the perfect fit

Zappos provides highly fluid and responsive customer experiences by using
analytics and machine learning built on AWS. These services enable Zappos to
personalize sizing and search results and to offer highly relevant and enticing
product recommendations based on individual user preferences.

Run, don't walk

Zappos delivers these personalized search results and recommendations with

nearly undetectable latency—99% of searches complete in less than 48 mil-
liseconds. These features have helped Zappos reduce repeated searches and
product returns while increasing search-to-product-clickthrough rates.

Read the story >

15
THE SUCCESSFUL ORGANIZATION

Serve all your customers’ needs

Organizing for success
It’s all about making a mental shift from:
Technology is only one component when building an advanced analytics
capability at your company. The organization—its processes, governance,
and people—creates the biggest challenge. So how do you organize your
analytics investment for success? “You must use our reporting solution
and you will like it.”
Start with an analytics Center of Excellence
Spearhead the change with a team and a leader as you move from strategy
and intent to meaningful progress. Set up an analytics Center of Excellence
(COE), starting small and growing to service more needs.

Large enterprises that already have established shared services organizations

for business intelligence or reporting can seed the analytics COE with techni-
cal and business roles. Similar to IT infrastructure organizations, they should “What are your analytics needs and
not only supply the talent, but also be a key driver and sponsor for the effort. how can we help enable you?”

Over time, they will evolve to either adapt or become part of the analytics
COE. The starter roles are often data engineers and business intelligence
analysts. The leader will work across multiple organizations, business units, When establishing a new analytics COE, set expectations for how
and back-office groups, such as finance and IT. the group acts and makes decisions. Position the shared services
reporting organizations to answer the different questions that
come from employees, business leads, and customers.

"It’s not about having the latest tools in the belt; it’s about
making it easy for your customers to get what they need."

16
TWO CUSTOMER TYPES

The analytics COE will need

to service two types of customers

The data and analytics consumers

The decision makers, data scientists, business intelligence (BI) analysts, and
developers who are concerned with the quality of the tools and services
available to them to process and present data.

The data producers

The owners of applications, infrastructure, and devices who will supply the
data into the platform. These customers need services to easily publish their
data into the analytics platform and define a data contract.

This includes the domain model of the data, frequency of refresh, and
definition of policies—for example, a security policy outlining who can access
their data.

Those who manage the data and stand between data producers and data
consumers often have a difficult mission. Data managers communicate across
a potentially very large and diverse set of business units and personas—and
if they do not meet expectations, then the analytics effort will not deliver the
business value.

17
MODERN ANALYTICS

Rethink the COE

An analytics Center of Excellence presents Adapting to evolving technologies

a specialized set of cloud services focused This is the era of rapidly evolving technologies, including big data, rich
visualizations, automated decision-making, AI, and machine learning.
on meeting analytics needs.
In today’s climate, it’s just not possible to have a single technology stack.
COEs run the risk of becoming concierge services. That can be fine for certain
types of requests, but the COE can quickly get overwhelmed and backlogged
with requests. There needs to be scalable, self-service mechanisms and
transparent prioritization and governance processes in place.
The past
Reporting and BI organizations often provided Analytics COEs need to engineer and architect data platforms that are
a one-size-fits-all strategy self-service, secure, operable, and scalable. The ecosystem needs to be ever
evolving with technologies that can process, analyze, and present insights.

The present
Making it easy for your customers (producers
or consumers) to get what they need

18
CLOSING SUMMARY

Sink or swim–What do I do today? The steps forward

Put your data to work. Move it to the cloud. 1. Establish a data lake as your central, all-encompassing
data repository in just a matter of days. Establish security,
Data is your strategic asset—have all your data, applications, and analytics governance, and auditing policies in one place, applicable
workloads in the cloud. Start with one project and, upon its success, gain across the data lake to all users and all applications.
momentum to move on to the next project and the next.

2.  tilize crawlers to crawl through your existing data silos

U
and create a centralized data catalog that can be used by
multiple services.
Benefits
Speed, agility, and flexibility 3. Connect
 your cloud data warehouse and other analytics
tools to the data lake and data catalog.

Avoid
Planning, buying, installing, 4. Empower business users, developers, and data scientists
to access the same data catalog and data. Enable self-
configuring, managing, and service to the data through multiple, fit-for-purpose
maintaining infrastructure analytics tools.

5. Keep up the momentum and focus. Maintain open and

frequent communication with your data and analytics
consumers and your data producers.

19
CLOSING SUMMARY

Data harnessing starts by working modern

A modern analytics platform, with a data lake at its center, opens up a world of
opportunities that can help your organization harness the power of data.

Imagine this journey…

…From the moment you begin to book your next trip, the airline, the hotel,
the car rental, your data can make exclusive choices just for you.

Your favorite airline seat. The car you love to drive by the beach. The hit
playlist, you know the one, that only you could find relaxing. Maybe it’s
the new summer read, for the beach only. That movie you didn't know you
wanted to see, delivered to you on board. Or that cheese you’ve always
Learn more about Data Lakes
wanted to try, in that perfect idyllic local restaurant, yes, it really does exist. and Analytics today >
With your data, discover the best adventures, perfect for you.

Kick-start your journey today

It all starts with capturing and storing data in a data lake and analyzing the
data with a modern analytics platform. The more data you collect and keep,
the greater the possibilities for using that data in new and innovative ways.

Plan out your analytics solutions, map out your data architecture to incorporate
data lakes, and harness the power of your data.

Today is the day to become a data-driven organization.

20