Informix Products: Informix-Online Dynamic Server Informix-MetaCube
KFC, the restaurant chain owned by giant
PepsiCo, recently implemented a datawarehouse based on technology from Informix and
Hewlett-Packard. The warehouse replaces a number of smaller data repositories with a
single, corporate-wide infrastructure and is expected to give KFC valuable information
about individual store performance, as well as overall customer buying practices and
product preferences-from both historical analysis and up-to-the-minute perspectives.
Getting to know your customers
“Everybody needs a little KFC,” goes the advertising slogan and,
apparently, it’s not far off the mark. KFC, PeopsiCo’s $7 billion restaurant chain,
operates close to 9,000 company-operated and franchised quick service restaurants in over
70 countries-2000 of those are company-owned in the US.
When you’re the world’s largest chicken
restaurant chain, your IT department constantly deals with immense volumes of data in
order to pinpoint customer preferences. For example, what are the most current customer
demographics? What are they eating-and at what time of day? How’s the new menu item
testing in target cities? Compared to each other, how are various KFC locations in any one
The need for a new framework
Over the years, KFC has conducted data analysis using a variety of data
repositories, front-end tools and disparate systems. The problems with the existing
framework were numerous. First, although there were many tools for getting answers from
various systems, none of them worked together to form a cohesive picture of the business.
For example, queries to determine sales had to be made to one system, while queries about
product inventories had to be made to another. “The need for data analysis kept
growing and growing,” explains Micki Thomas, Systems Development Manager for KFC’s
Restaurant Support Center. “We didn’t have a solid infrastructure in place to handle
the demand for information.”
Michelle Wells, Director IT KFC, agrees.
“We had to get to a level of detail that the systems we had in place could not
handle, including customer transaction levels by time of day and product, for all of our
domestic restaurants. We had it for a small subset of restaurants, but we needed them
all-which required a fundamental shift in the infrastructure to support the amount of
information we wanted to store.”
KFC not only had to make all of the
information available, but it also had to be easy to use. “We’re talking about
strategic planners who should be spending 80% of their time figuring out what to do with
the business, not plugging in numbers and figuring out spreadsheets,” Thomas says.
“We wanted users with minimal training to do a lot of the reporting themselves. We
didn’t want to have programmers writing queries for them all day long.” Ultimately,
KFC decided to build a datawarehouse with the flexibility, scalability and data capacity
necessary to not only replace all of the existing systems, but to grow as information is
added from stores on a daily basis.
Planning the perfect datawarehouse
KFC’s datawarehouse project is a model of requirements analysis, thorough
planning and subsequent execution. “We kicked off the project with a joint
application design that a consulting company helped formulate,” explains Wells.
“We included our IT organization, key clients in various corporate departments and
people in the field offices in the process.”
“We started out by trying to
understand all of the potential client demands for a new decision-support system. Those of
us who had been involved in putting together our existing systems determined requirements
from a technology stand point-what we needed today, as well as what we would need in the
future.” The inevitability of change was a constant theme for the group, which knew
that as internal customers come and go, the demands on the system will change accordingly.
With that in mind, KFC focused on finding datawarehouse technology that could shift as
quickly as might be necessary.
The requirements formulated by the Restaurant Support Center’s project team were
that the system be open, scalable, affordable and flexible. Says Well, “Somewhere
down the road, PepsiCo’s other restaurant chains, such as Taco Bell, may want to use the
system-so we didn’t want to do anything they can’t leverage.”
The team also concluded that the system
needed a consistent front-end with an ability to customize it, from an easy, intuitive
interface supplied to most users, to powerful, custom-query environments and add-on tools
that sophisticated users require.
Users told the team their most basic need
was for data to be both more timely and more accessible. They wanted ‘one-stop shopping’
for data-one place to drill up or down for as much data as they required. And they wanted
to be able to run queries on background mode, leaving their terminals free to work at
other projects or tasks.
Finally, the datawarehouse itself needed to
handle multidimensional data analysis, incremental data loading, client meta data
management, data security and ongoing performance monitoring and tuning.
In order to meet these diverse requirements, KFC chose Informix-Online Dynamic
Server, based on Dynamic Scalable Architecture (DSA), the parallel database architecture
that is at the heart of Informix’s database server product line. DSA delivers unmatched
performance, scalability and speed in enterprise-wide client server environments.
Online Dynamic Server is running on a
four-processor Hewlett-Packard 9000 series hardware platform. KFC is also using an array
of relational online analytical processing (ROLAP) tools, including:
– Informix-MetaCube Explorer
– Informix-MetaCube Warehouse Manager
– Informix-MetaCube QueryBack and
– Informix-MetaCube Aggregator
With the combination of these technologies,
KFC is equipped with better customer information, which ultimately makes the company more
Putting the warehouse in place
Currently one year of data has been loaded into the datawarehouse. Eventually,
six years of data will be housed in it. As KFC staff populates the warehouse, its eventual
size is an ongoing topic of discussion. “Figuring out where it’s going to go is
probably one of the biggest challenges of all,” Thomas says. “Eventually, we’ll
probably be at 200- plus GB-so system scalability is critical.”