SPECIAL SECTION: Oracle
at C-Level DATA MANAGEMENT
By Louise Fickel
Sinking under an ocean of
data? Try these strategies.
If you're a ship's captain
today, you can get better data sooner about the waters that lie ahead, but
the seas remain the same finite entity Magellan faced in the sixteenth
century. If you're an airline pilot, the same is true: Information about the
skies in front of you comes in more quickly now—and more accurately—but Earth
is still the same 24,000 miles in circumference it was when the Wright
Brothers first took off at Kitty Hawk. But if you're a C-level executive and
the vessel you're piloting is a global enterprise (or aspires to be one), the
amount of data now before you seems to have conspired to become the Perfect
Storm—a deluge of facts and figures that, if not properly managed, may be
enough to sink the biggest of ships. The question for many executives is no
longer, "How do I get the information I need?" but rather,
"How do I make sense of it all, and where do I keep it,
Regardless of where you are
in a discussion that likely now includes terms such as "direct-attached
storage devices" and "networked storage devices," you will
probably return again and again to the topic of data management. "What's
happening now," says Don Swatik, vice president of alliances and
information sciences at Hopkins, Massachusetts-based EMC, the world's largest
manufacturer of data storage devices, "is an evolution from an
environment where you have one server connected directly to one unit of
storage with very little management to a more-sophisticated environment of
multiple servers from different vendors, heterogeneous operating systems, and
multiple storage units from multiple vendors. Now you need to manage in three
different spaces of server, network, and storage. So it becomes a question of
how to simplify, automate, and make it open."
Storage management is at Oracle's
core. With Oracle's new online data-reorganization capabilities, database
administrators can move tables and indexes from one storage device to
another without taking the database offline. Oracle Recovery Manager,
meanwhile, handles backup and recovery for multiple databases. Learn more
Yet if you want to manage
your data effectively, you must first design a storage architecture that
ensures your data is consistent, accessible, and can be analyzed across the
enterprise. Consider this: Many companies now store their customer
information in multiple formats in isolated silos across the company. The
call center, for example, defines John Smith as "Smith, J.," while
the marketing department identifies him as "John Smith Jr." In this
same scenario, rather than various departments dipping into the same pool of
information, employees access different islands of data. The result is that
everyone has a slightly different picture—and understanding—of the same
customer. Not good for the customer. Not good for you.
To develop a coherent
storage architecture, your management team must work together to first define
your company's strategic goals and then determine the types of data needed to
support that business strategy. "For many organizations, it's difficult
to define the kinds of information you need to control the company,"
says Herman Verpalen, manager infrastructure architecture and management
information at the Netherlands division of pharmaceutical giant
GlaxoSmithKline. "The real issue is to help the management team discover
its own business and ensure that you have the correct business
Once you have a clear understanding
of the business, then you can determine who needs access to which kinds of
data and how the data should be linked. The next step: implement
storage-management tools from vendors such as EMC, Hewlett-Packard, Hitachi
Data Systems, Legato, Network Appliance, Sun Microsystems, or Veritas to
automate and streamline tasks such as switching storage from one machine to
another when capacity starts to run low on the first one.
Close and Personal
But how does all of this
play out in the real world? By mapping out comprehensive strategies for
storing and managing their data, three Oracle customers are now using their
data to develop stronger customer relationships, respond more quickly to
changes in the marketplace, and run their businesses more strategically.
GlaxoSmithKline, which sells
nearly US$30 billion worth of prescribed and over-the-counter medications
annually, keeps its finger on the pulse of markets in 150 countries worldwide
by running a network of astute local sales and marketing organizations. At
the heart of GlaxoSmithKline's data challenge: Products are distributed
through doctors and hospitals rather than directly to patients, so making
sense of sales data is difficult.
Two-thirds of IT
professionals said storage issues are now discussed strategically within
their companies. More than half mentioned that senior management now pays
more attention to the topic. Top four reasons cited for a growing demand
for storage: business continuity, ERP, CRM, and internet applications.
Source: Lorraine Cosgrove
Ware, CIO magazine
In the Netherlands,
GlaxoSmithKline employees track local market conditions and assess the
effectiveness of their own marketing strategies by analyzing huge volumes of sales
data on the company's own products and those of competitors. A centralized
Oracle data warehouse—which is expected to grow by 6 million rows per
month—consolidates, stores, and analyzes point-of-sale data gathered from
numerous sources, and the center's staff of 500 accesses the data via four
online data marts. "We made all our data open and available to everyone
because we wanted to ensure transparency across the organization," says
GlaxoSmithKline's Verpalen. "We have data objects that can limit access,
but that's never been necessary. Within the next year, we plan to include
other information, such as cost and action data, and add more business
intelligence so we can track return on investment (ROI) more
Flexibility was crucial in
developing the IT architecture. "We needed the flexibility to respond to
business changes, especially in light of last year's merger between
GlaxoWellcome and SmithKlineBeecham," says Verpalen. "Because of
Oracle's performance, reliability, and strong analytical capabilities,
extraction, transformation, and loading processing of our data warehouse is
fast, efficient, and easy to manage. Although a lot of data warehouses aren't
very flexible, we have a fast, manageable architecture. With Oracle and other
components of the architecture such as mnidex and BusinessObjects, we can
remodel quickly and adapt to business changes within days."
By all accounts, the Dutch
organization is already off to a good start. "Based on what we're seeing
in the marketplace via the data warehouse, our executives have decided to
reorganize the entire operation so we are more closely aligned with our
parent company's strategic goals," Verpalen adds.
Reality of Retail
And what about retail
organizations whose annual revenues are typically generated within an
absurdly short period of time? Nokian Tyres, a tire manufacturer and retailer
based in Finland, knows all about small windows of opportunity. The company,
which specializes in winter tires for cars and vans, sells most of its product
during the first three weeks of November. Yet throughout the year, the
company produces 5 million tires and must be able to track about 1 million
units of inventory on any given day of the year.
Nokian uses Oracle for its
data warehouse and its enterprise resource planning (ERP) system to support
the retail and manufacturing sides of the business. "Our data warehouse
is a consolidation environment that supports high-level business
decisions," says Raimo Mansikkaoja, vice president of IT at Nokian.
"We consolidated all our logistics and sales-reporting data last year so
we could manage it more easily and track inventory more effectively."
People throughout Nokian—from executives and the sales force to manufacturing
and R&D—can access data according to job description. As at
GlaxoSmithKline, backups are completed automatically with software. And
everything is done on a tight budget: 16 people and an annual IT budget of
about US$5 million.
Nokian has managed its
technology creatively. "The role of our data warehouse has grown
significantly over the past year," says Mansikkaoja. "We have taken
nearly everything away from the operational environment so we can meet our
data requirements more easily, faster, and at a lower cost. Because we've been
able to eliminate most of the reports produced directly from our ERP
environment, our end users who are waiting for those reports now get better
service. They use Oracle Discoverer on their PCs to access the data warehouse
directly. "From a management point of view, we can now support our
decision-making more effectively," he adds. "For example, we can
combine bill-of-materials data with order-stock and manufacturing data
instantly. That solution is also good from an IT point of view, because we
reduce the ERP load. In turn, we have been able to dedicate more capacity for
transactional processing and reporting during the high season. Oracle has
helped us see our business processes more clearly and develop them in a
different way so we can run the business more efficiently."
Outsourcing also plays an
important role in Nokian's strategy. "We outsourced our Microsoft
Windows NT servers and workstations to concentrate on our core
business," says Mansikkaoja. "We know our ERP system and data
warehouse well, so we are managing those systems ourselves. Someday, I
expect, we will entrust our mission-critical applications to service
providers. But before we do that, we want to make sure those experts are
available to us 24 hours a day, especially during high season."
The question of data storage
and management will undoubtedly continue to come up. "You can't sit back
and wish that the world were 'vanilla' and hope the complexity will go
away," says EMC's Swatik. "The biggest mistake a company can make
is to implement an infrastructure that doesn't scale into a more complicated
environment down the road. As databases continue to increase by 50 to 100
percent each year and management problems grow geometrically, the only way to
solve the problem is by networking the information and using sophisticated,
automated management so that, in the future, one person can manage 10 times
the amount of data they can manage today."
Louise Fickel writes about technology and
business for industry publications such as CIO. Based near Washington,
D.C., she can be reached at firstname.lastname@example.org.
Drowning in Data?
Strategies for Keeping Afloat
How can you begin thinking
strategically about data storage and management? Experts point to a variety
of business and technology issues to consider:
Consolidate Consolidating your data offers
several distinct advantages: You can keep it more secure in terms of access
and disaster recovery. Management is easier and more cost-effective.
Centralized, networked data also facilitates the flow of information.
"In the application space, data in one database is typically dependent
on data in multiple databases," says Don Swatik, vice president of
alliances and information sciences at data-storage manufacturer EMC.
"If you need the ability to share information across applications,
networked storage ensures better performance because the data doesn't have
to go through the local-area network (LAN) first."
Segregate Determine who should have access
to what data depending on the particular business issues that define your
company. However, some general business rules apply across industries. Ask
how you want to use the information. Keep in mind not only your business
drivers but also the size and dimension of your business and how the data
will be integrated with core systems. Define the logical and physical model
of your data warehouse with your customers in mind if you want to
distribute your customer information across the company, for example.
Scale Capacity and scalability are also
key issues. "Companies should look at the amount of data they expect
to have, the number of users, and the response time their users can
tolerate," says Cyrus Shahabi, assistant professor of computer science
at the University of Southern California (USC) in Los Angeles. "With
those performance metrics, they should ensure their system can handle the
throughput and desired latency time." If you're not familiar with
determining this kind of information, talk with someone on your IT staff. Even
if staff members don't know exactly how much data your sales organization
creates every month, they can at least generate histories to provide a
springboard for discussion.
Cost Cost of administration is also
important. For example, finding even one person who can administer a
heterogeneous environment is sometimes difficult—and expensive. If total
cost of ownership isn't considered, your project won't be successful.
Technology You'll also want to spend some
time understanding the nuances of specific technologies. For example,
direct-access storage devices—first-generation storage products that
connect directly to servers via network interface cards—may be more useful
for storage and management of data on Web and file servers. "They
modify the operating system and throw away most processes not required for
data management to improve the data-intensive processes, so they're very
efficient in pumping input and output," explains USC's Shahabi.
"They are not good, however, for database applications, because
databases do lots of optimizations that aren't compatible with the OS
optimizations of network-attached storage devices. So storage-area networks
are better for databases."