9
EBOOK Dan Brock Head of Customer Success Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball A New Non-dimensional Approach to Data Modeling and 3 Ways It Can Transform Your Organization

EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

  • Upload
    others

  • View
    46

  • Download
    5

Embed Size (px)

Citation preview

Page 1: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

EBOOK

Dan BrockHead of Customer Success

Death of A Star Schema (Redux): Moving Beyond Inmon and KimballA New Non-dimensional Approach to Data Modeling and 3 Ways It Can Transform Your Organization

Page 2: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

2

In 1990, Bill Inmon introduced us to a “top-down” approach to managing data in the enterprise, starting

with a third normal form (3NF) data warehouse that encompassed all of a company’s data. From there,

data was pushed to cubes or data marts as needed, as the original 3NF enterprise data warehouse could

not perform well enough for analytics and visualization to be run against it. The initial investment of

money and man-hours was daunting, but the result was a holistic data view of the entire enterprise that

enabled a wider range of questions to be asked of the data.

Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that data warehouses

should be easy to understand and fast to respond. In other words, by defining the questions that needed to

be answered and insight that needed to be drawn from the data, a company could architect more efficient,

better-performing data marts using wholly dimensional data (where Inmon would only do so in individual

data marts). This approach worked for many, and thus began the rise of our old friend the star schema,

whose popularity continues today, often to the annoyance of data professionals across the industry.

Picking the right data model is one of the most important decisions that senior IT leaders can make, with

the vast majority historically defaulting to a “bottom-up” approach to mapping atomic data. Recently,

however, some of the traditional limitations to “top-down data” modeling are gone, leading to a limited

resurgence of Inmon’s model.

As we discussed in the first installment of Incorta’s Death of the Star Schema series, the technology that

supports BI and data warehousing has evolved rapidly in recent years. Now, there is a third option for

data warehousing and BI in a post-star-schema, post-ETL world: Non-dimensional data modeling.

What’s Happening

Inmon's top-downapproach

Kimball's bottom-up approach

Sources

STARETL Report

3NF (Atomic)ETL Cube Mart Report

3NF

Report

TransformTransformBy connecting directly to data sources, Incorta is able to speed up implementation, deliver high performance, and give a complete, top line to transaction view of the data.

Page 3: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

3

Incorta’s Direct Data Mapping eliminates the traditional need for data transformations assumed

necessary by many, returning to a non-dimensional 3NF data warehouse, where transforms can be applied

as needed and analytics and visualizations can be run directly against the data faster than either of these

traditional architectures. Complex queries that once took hours can now be completed in seconds, and IT

can deliver fast and flexible self-service BI that works for the entire organization.

What does this mean for IT teams? In this latest

installment of the Death of the Star Schema

series, we revisit the age-old data modeling

debate and seek to move past it with a novel

vision of how IT leaders can make a new data

reality for their business. We examine the

benefits of this new way of thinking about data

for IT teams and business users, and explore

how a company’s approach to data will change

without the need for ETLs and star schemas. It’s

no longer a choice between top-down, bottom-up,

or something in between.

Inmon Model Kimball Model Incorta

Performance Average Average Great

Refresh Rate Slowest Slow Near-Real Time

Fidelity Good Poor Best

Agility Average Poor Best

Initial Deployment Slowest Slow Fast

Slowly Changing Dimensions N/A Yes Yes

Cost High Average Best

There is now a third option for data warehousing and BI in a post-star-schema, post-ETL world: non-dimensional data modeling.

From Inmon to Kimball to Incorta. Based on modern data principles, Incorta offers unmatched performance without the implementation overhead.

Page 4: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

4

Asking the Right Questions About Questions

Kimball’s bottom-up approach to data modeling has been the go-to strategy for data-driven

organizations for decades. The targeted approach and fast time-to-value made it a natural choice for

companies experimenting with data for the first time, especially the early pioneers.

Until recently, the prescriptive nature of Kimball’s model was an advantage for organizations looking to

glean more insight from their data. They identified the key business questions that needed answering

and set up data warehouses that would be able to answer those questions. It was a relatively simple and

direct approach—and more cost effective, too.

The shortfalls of Kimball’s approach, however, are

becoming increasingly apparent as data technology

advances. The very strategy that makes bottom-

up data structures so efficient—preordaining the

questions to answer and problems to solve—is also

extremely limiting. That is to say, when you restrict

the search area for business questions, it robs you

of potentially game-changing insights that come

from unexpected sources. You lose the ability to

ask follow-up questions, ask adjacent questions,

and deviate from the prebuilt path. That’s a big

deal because unexpected insights are often the

most impactful for a company.

What Inmon realized early on is that a 3NF representation of a company’s data best equips that

company to retain the fidelity of the data and, thus, answer almost any question of that data the

business may have. The rub, however, is performance. While Inmon’s vision for a holistic, top-down

approach to data modeling is easier to achieve today than ever before, performance issues remain

because the model requires data to be pushed to cubes and marts.

This “question of questions” is at the very heart of the philosophy underpinning data design. In

today’s world, seeking tools to answer specific questions is the wrong approach. What’s needed are

tools that allow us to freely ask and answer questions with data. And work has already started on tools

that answer questions we haven’t even thought to ask yet. Now that is business intelligence.

Ask yourself what it is, exactly, that your company wants to ask itself. There are basic questions you

will always need to answer, of course—but there are also “unknown unknowns,” which, by their nature,

cannot be predicted. Incorta’s approach opens up the possibility of identifying the unknown unknowns

in your business, and allows your company to approach BI in a way that more accurately reflects the

real world of business. Innovation and disruption often come from unexpected sources—equip your

company accordingly.

“Incorta’s approach opens up the possibility to identify the unknown unknowns, and allows your company to approach BI in a way more accurately reflects the real world of business: innovation and disruption necessarily comes from unexpected sources.”

Page 5: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

5

Laying a Foundation for Data Curiosity

Now that traditional constraints to data modeling are disappearing and third path has been cleared, IT

Leaders need to update their fundamental expectations about what data can deliver.

Kimball’s approach is point solution. It can help

you answer specific types of questions with data—

many of which are highly valuable—but that’s

about it. Inmon’s top-down approach, on the other

hand, can be the start of a broad organizational

cultural change in the way data is used and insights

are uncovered across the entire organization.

Inmon’s approach begins with constructing the

master data warehouse, a centralized store of

all data available from across the business. From

there, data marts are built as needed. But while

this greatly improves the breadth of questions that

can be asked, IT Leaders often find the enterprise

3NF warehouse to be lacking in speed and agility.

Now there exists a solution that combines the strengths of these two models and overcomes their

limitations. With Incorta’s Direct Data Mapping, the virtues of both approaches are combined and

accelerated. We have access to data at an enterprise level and the ability to query against it freely and

flexibly while still getting answers faster than ever before. For IT Leaders, the difference in scope and

vision is immediately apparent. No longer are lines of questioning constrained, nor are answers segmented

department by department. Your vantage point over the entire company evolves from being a patchwork of

department-level insights into a connected, enterprise-level view.

This can be the foundational step in creating and enabling a culture of data curiosity. As business becomes

digitized and interconnected, IT leaders in every industry are finding that problems are more horizontal

than vertical, and that modern BI can tell a broader story about an organization than it ever has before.

For example, a problem for the sales team is most likely going to be a problem for the finance team as

well. Likewise, the answer to a question emerging from IT may lead to a solution to a seemingly unrelated

problem in HR.

We talk a lot about fostering a culture of “data curiosity,” in which employees across the company are

empowered and enthusiastic about self-service BI. That’s because it’s good for business and people

inherently want to be more data-driven. But it only works when they have the ability to ask the questions

“There’s no division within [our organization] that Incorta hasn’t touched.”—Business Intelligence and Analytics Director at a Leading POS Provider

With Incorta’s Direct Data Mapping, lines of questioning are no longer constrained, nor are answers segmented department by department. Your vantage point over the entire company evolves from being a patchwork of department-level insights into a connected, enterprise-level view.

Page 6: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

6

Recasting the Role of IT Leaders

This renaissance of reduced data modeling is recasting the role of IT Leaders in multiple ways.

There are no shortage of IT Leaders with shining track records of scoring “quick wins” on data and BI

through bottom-up approaches. These short-term victories are important and propel many companies

out of the doldrums or into their own golden eras. But the never-ending proliferation of data and its

soaring value demand that IT Leaders take a long-term view of the data, the company, and

the industry.

Inmon’s top-down approach has generally been regarded the stronger long-term option, and was

important in forcing a change in the relationship IT Leaders have with their data and how they provide

BI to their organization. But it remains burdened by its clumsiness and is still reliant on the additional

step of cubes and data marts, further increasing the time it takes to arrive at actionable insight.

As we discussed in our recent eBook, Avoiding Self-Service BI Disasters, shifting your mindset from

“control” to “empowerment” is key. Your job is no longer to limit or control the access that departments

have to data but rather to enable and empower them with data of the highest fidelity that they can

confidently use to become “data scientists” in their own right.

With Incorta, IT Leader thinking moves beyond the

short term—endless requests for new dashboards,

catching and correcting bad data, and spinning up

ad hoc data warehouses of questionable fidelity.

Going forward, the questions you need to ask are

bigger and more strategic in nature. For example:

How can we enable self-service BI and foster a

culture of data curiosity? How can we ensure

accurate data in a self-service BI environment?

How do we quickly incorporate new data sources

without elaborate data modeling for the sake of

improving query performance? How does the

role of the data team evolve as data becomes

democratized?

they want to ask, when and how they want to ask them. Confining the scope of those questions or limiting

insight on a departmental basis— the Kimball model—runs antithetical to this philosophy. Inmon’s approach

lacks the speed and agility to keep pace with modern business. Incorta improves on both and can catalyze a

new way of thinking at a company-wide level.

There are no shortage of IT leaders with shining track records of scoring “quick wins” on data and BI. But the never-ending proliferation of data and its soaring value demand that IT leaders take a long-term view of the data, the company, and the industry.

Page 7: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

7

Incorta: Moving the Discussion Beyond Inmon and Kimball

Incorta challenges decades of assumptions about the best way to architect data in the enterprise. Ralph

Kimball’s approach has been popular for its speed and low cost to entry, and recent advances in data

technology have made Bill Inmon’s richer and more extensible top-down model more viable.

Their fundamental limitations, however, persist. Only by changing the rules once considered set in

stone—the need for endless ETL processes, the star schema, data cubes and marts—can we make

meaningful, non-incremental steps towards the future. Incorta and its Direct Data Mapping capabilities

finally allow us to step outside the borders of the canvas and introduces an entirely new option that is

no longer tethered to the limitations of traditional data infrastructure.

This shift impacts the long-term role and responsibilities of IT leaders, whether they are a VP, CIO,

or CDO. It also lays the groundwork for a cultural shift in companies that endeavor to empower all

employees to become data scientists.

Tackling the Unknown Unknowns

A bottom-up approach to data modeling

requires many assumptions about the

problems you are looking to solve in the

business. While this is no doubt a useful

approach for well-defined problems,

disruptive innovation usually lies beyond

the domain of the everyday. With Incorta’s

Direct Data Mapping, IT leaders can take a

holistic view of data at the organizational

level and open the doors to insights that

are not bound by assumptions. By making

data available without pre-supposing a

model of the questions a business might

ask, Incorta elevates Inmon’s vision into an

even more impactful model for businesses.

Page 8: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

8

Enabling Data Curiosity

Incorta can unlock data across your

organization. What was once an unwieldy,

time-consuming process can now be mined as

a source of new and business-changing insight.

In the bottom-up method of doing things,

natural delineations between departments

and disparate data warehouses naturally arise,

pushing your company’s best thinkers into

silos. Designing and constructing a top-down

data architecture can be so cumbersome as

to set you back, and requires diligent, timely

maintenance. An enterprise-wide, agile, and

fast 3NF warehouse enabled by Incorta is a

foundational step in enabling data curiosity and

individual data exploration at your company,

setting the stage for disruptive insights from

every corner of your organization.

Evolving IT Team Roles

In many ways, the importance of IT leaders has

lagged behind the importance of the data itself.

With Incorta’s Direct Data Mapping, IT leaders

can take the first steps toward freeing themselves

from the everyday struggles of providing

compartmentalized analytics and BI, and begin

taking a longer-term view. Incorta’s Direct Data

Mapping can change the way data—and the teams

that supply and maintain it—are regarded within

your company.

Without question, Bill Inmon and Ralph Kimball

remain two of the greatest data thinkers in our

nascent field. Their approaches have been the

starting points for just about every company

trying to better themselves through smart

usage of data, analytics, and BI. Kimball’s design

has become the default, and as technology

has evolved, the viability and usefulness of Inmon’s top-down approach has grown. But if long-term

strategic thinking built on data and analytics is your goal, perhaps it is time to consider what comes

after these two models, each of which is still saddled by their core assumptions. Technologies like

Incorta and its 3NF warehouse can enable a completely new view of the world of data.

Page 9: EBOOK Death of A Star Schema (Redux): Moving Beyond Inmon and Kimball · 2020-02-24 · Not long after, Ralph Kimball evolved his “bottom-up” approach, led by a basic belief that

To learn the fastest way to what matters, visit www.incorta.com or join the conversation on Twitter @incorta.