Page 13
Page 12
Data Warehouse is simply a database. But what
makes it different from transactional Database is
because it’s designed for analytical purpose. A good
Transactional Database shall be normalized into
separate relational tables (to reduce redundancy),
but this condition can become a disadvantage when
we want to analyze the data by querying many re-
lational tables. The query will take much resources,
as well as a longer time to finish (especially when
you want to build dynamic report withmany dimen-
sions *read it: Business Intelligence).
So, a man named William Harvey Inmon came up
with an idea to rebuild the structure of the database
by denormalizing the relational tables into a simple
schema called Star Schema. The star schema itself
is later called the Data Warehouse, and the process
to get the data from Transactional Database, denor-
malize the data, clean and harmonize it, and then
insert it on Star Schema’s Tables called ETL Process
(Extract, Transform, Loading).
Data Warehousing, as a practice, encompasses a
much larger scope than a simple database. That’s
where it starts, but to be of full value, processes such
as Data governance and meta and master data man-
agement are essential. Data Warehouses are typi-
cally used in two capacities, either as a centralized
hub for integrating applications in need of common
data or as a repository for data to be used in analyt-
ics and reporting.
Data modeling is usually broken into two camps
of thought: the Inmon or Kimball approach. Inmon
favors a totally third normal form of the data and
created the Information Factory approach, which is
a single version of the truth and promotes a single
relational model for all data. Kimball was the individ-
ual actually responsible for the proliferation of the
star and snowflake schemas, and the uses of data
marts which provide specialized versions of data for
reporting and analytical usage.
Inmon is known as the father of data warehousing
for coining the term, while Kimball is known as the
father of business intelligence. Most data warehous-
es use a hybrid model of a third normal form in a star
or snowflake schema.
What Is The Market Size For
Data Management Platforms?
What Is A Data Warehouse?
As far as DMPs, there is only a handful that I would
call true DMPs (i.e., Adobe, Krux, Bluekai, and Lotame).
Then there are several pseudo-DMPs, usually compa-
nies that specialize in another sector of online adver-
tising (e.g., main product being a DSP or tag man-
agement, etc.), such as Audience Science. On top of
that, there are several networks and even individual
publishers out there which have created their own
internal DMP for their personal use and that of their
clients.
As far as demand goes, it’s increasing rapidly—and
has been since the beginning of 2012, and will prob-
ably only increase at a faster rate within the coming
year. Value can always be found in DMPs if one uses
it for the right reason(s), regardless of the size of the
individual publisher, marketer, network, etc. Even
those companies that get less than 1 million monthly
uniques can still find value by selling their data in
DMPs at no cost to them for setting this up. I don’t
see demand ever being an issue in this industry going
forward.
As far as countries go, the US is obviously further
along in the DMP hunt than elsewhere. Europe (as
well as most of the other continents), I’d say, is where
the US was a year ago. Companies are more wary of
signing up with DMPs there, as privacy is still a major
concern for companies (unlike in the US, where it’s
treated more as a major nuisance, so it seems to me).
Another reason the US is further along in this transi-
tion is due to third-party data. Most of the major data
providers that have actually on-boarded data online
are located in the US. We are seeing more and more
providers elsewhere, Europe, for example, but this
process can be slower than some would-be DMP cli-
ents might like. The same can be said of on-boarding,
these would-be clients’ own offline data.
Regardless of the stage the DMP industry is at in the
given country, it will be increasing in size at a tremen-
dous rate in the near future.
