|
It's time to develop a strategy
The future is never certain, but there
is one thing you can count on—the concept of a real-time
enterprise is not going away.
by Stephen Poole
 s
data warehousing enters its third decade, the phrase "real-time
enterprise" has become something of a mantra in the industry,
and for good reason. Success in the rapidly shifting 21st-century
economy will hinge on a business's ability to identify and
respond quickly to emerging trends, shorten development and production
cycles and streamline operating costs while simultaneously minimizing
business risks—precisely the capabilities that a properly
implemented enterprise data warehouse can provide.
Cynics who think "real-time enterprise"
is merely a buzzphrase designed to encourage wanton IT expenditures
should consider this: Gartner analysts have predicted that in
just two to three years, the dominant players in every business
segment will have achieved leadership or solidified that position
through real-time capabilities (www.gartner.com).
But there's one small fly in the real-time
enterprise ointment: The group playing one of the most critical
roles in bringing real-time interaction to data warehouses—IT
managers—seems to be the one least convinced it can be achieved.
A January 2003 Gartner survey of 120 IT managers
from mid-size and large enterprises in North America and Europe
revealed that a mere 6% believed that their business systems and
processes could be transitioned to real time by the end of the
decade.
 |
Todd Walter (left), CTO of the Teradata Development Division, and Stephen Brobst, CTO of Teradata, a division of NCR.
|
What is it that IT managers see-or fail to see-on
the horizon that justifies this level of apprehension? Is their
anxiety, to paraphrase media expert and author Marshall McLuhan,
the result of trying to do today's jobs with yesterday's tools?
Or is it indicative of a communication gap between the business
and IT sectors of many enterprises?
According to Stephen Brobst and Todd Walter,
Teradata's chief technology officers, it's a combination of both-the
good news is that these problems are not only solvable, but they
can also serve as launch pads to greater innovation and productivity.
And both agree with Gartner's assessment that the time to act
is now.
"If I were the CIO of a company that hasn't
begun to make the leap to a real-time enterprise paradigm," says
Walter, "I'd be scared to death right now. I'd know that either
the business process is going to change around me and I wouldn't
be able to keep up technologically, or I'd realize that I know
what to do with the technology but the business isn't making the
conceptual leap."
Bigger, faster ... and cheaper?
Perhaps one of the things that has the folks in IT a bit leery
are the numbers. While it's natural to expect a data warehouse
to grow in size as time elapses, the rate of growth can come as
a shock: A February 2002 Information Week article estimated that
the amount of data the average business collects and stores is
doubling every year. There are a variety of factors accelerating
that pace.
Increasingly complex regulations and laws have
forced businesses to save more data in order not only to prove
compliance but also to defend against potential lawsuits-a situation
that Walter says will "get a lot worse before it gets better."
Another is the introduction of Radio Frequency
Identification (RFID) for inventory tracking and management, which
will undoubtedly generate staggering amounts of data that must
be accessible in real time or near real time to be of any serious
value. And the growth of wireless communications, coupled with
the falling prices of microsensors capable of monitoring everything
from the reflectivity of road signs to the stress of individual
engine components, has the potential to generate mind-boggling
quantities of information. Given all that, it's not surprising
to learn that petabyte data warehouses are expected to go online
in the immediate future.
Can data storage technology keep up? Yes and
no, says Brobst. "Most people think about Moore's Law as it applies
to CPUs-computing speed doubles every 18 months. But the same
sort of phenomenon is going on with storage technology: Bit density
on the storage media is being doubled, except it's happening even
faster than every 18 months."
On the surface, it's good that data storage
costs are plummeting-from around $15 per megabyte a dozen years
ago to pennies per megabyte today. But the hidden cost is the
speed at which data is delivered from the disk spindles. That
speed is not being increased proportional to the increased in
storage space.
According to Brobst, each time a new generation
of disk drives comes out the performance will decrease up to 40%
if the same data is stored on the new technology. (For more on
this subject, read Ron
Yellin's article.)
What about transferring historical or archival
data from the data warehouse to other storage media-tape drives,
DVDs and so forth-to ease the burden? "A lot of people say that's
exactly the thing to do: Put older data on a tape or on a DVD,"
says Walter. "But what they don't consider is how often the data
might need to be queried and how long it'll take to read the tapes
or discs."
A phone company doing business in several states,
for example, might get five or 10 subpoenas a day for phone records.
Company personnel can't load the tapes each time and go through
years' worth of data to get answers. An enterprise may have this
history, but it doesn't know which particular test it will be
interested in tomorrow. So the access doesn't have to be lightning-fast,
but it does have to be efficient, according to Brobst.
The challenge, then, is to find a way to keep
all business data-historical records for strategic analysis, current
data for tactical queries-in the data warehouse while maintaining
the same level of performance. "The onus of responsibility is
on us here at Teradata to 'cool down' the disk drives by designing
software that doesn't require them to do as much work," Brobst
says. "That means introducing new techniques for indexing and
partitioning, (and) more intelligent I/O mechanisms, so that as
our customers drive toward more data on fewer spindles they don't
have to endure the accompanying performance decreases."
Teradata Warehouse 7.1 achieves this goal through
sophisticated allocation of processing and I/O resources. Based
on the relative service-level agreements for performance, the
software dynamically balances the data warehousing workload, with
tactical requests typically given priority over strategic queries.
In this way, a customer rep's request for a purchase history won't
be bogged down by someone analyzing inventory overhead for the
last four years.
The beauty of this design is that it does not
compromise real-time tactical needs at the expense of an in-depth
study of historical data, which in turn lays the bedrock for determining
the thresholds that will form the basis of predictive models and
automated business processes on which the real-time enterprise
will be based. (For more on this subject, see Carrie
Ballinger's article.)
Need an example of how critical the functionality
of efficient access to historical data can be? Just check out
the economic landscape of the last five years. "We haven't seen
this current kind of economic cycle in 10 years. Most people don't
have 10 years' worth of data in their warehouse … and now they're
finding they need it," notes Brobst. "Having more history allows
you to see across a range of economic cycles and be much more
predictive in your planning."
Realistic real-time
But don't let this lull you into a belief that simply being able
to perform previously time-consuming functions much more rapidly
is the final ingredient to the real-time enterprise formula. As
Walter quips, "Real-time enterprise without real-time data simply
speeds up old processes." Of course, "real-time data" doesn't
necessarily mean instantaneous acquisition and analysis of data
(though it might at some point in the future). For now, "real-time"
can be interpreted as "just in time"; that is, having access to
all the data you need, whenever you might need it.
The hard truth is that today's economy requires
new processes as it moves from a "push-based" to a "pull-based"
model, with the active data warehouse enabling businesses to adjust
rapidly to new demands and needs. "Dell is a perfect example of
this paradigm shift-they don't build a computer until they have
the order in hand," says Brobst. They do that not by maintaining
a massive inventory of components or by manufacturing components
themselves; instead, they "virtually integrate" their suppliers
through Internet technology and information-sharing techniques,
then use the data warehouse to monitor inventory and set production
priorities.
This stands in stark contrast to a push-based
economic model where sales forecasts done months in advance determined
base inventory, supply, production and hiring, all which works-until
something unexpected happens. At that point, the only option is
to "push harder": more advertising, more marketing, slashing prices.
That's not a desirable modus operandi in an economy where the
one thing you can expect is the unexpected.
Will it be easy for large companies to shift
to this new business model? Of course, the answer is no, but the
good news is that the switch to a real-time enterprise involves
evolution, not revolution. By making small adjustments in key
areas of the enterprise now, the cornerstone will be laid to embrace
new technologies and processes. And the very first step companies
should take, say Brobst and Walter, is to realize that the real-time
enterprise design encompasses the entire business.
"There are just too many organizations where
the IT and business departments are two completely different entities
that don't talk to each other and don't work with each other,"
says Walter. "IT departments are building technologies or rolling
out projects that are very tactical without considering all the
downstream implications or necessary business changes, or the
business is making changes without the IT guys putting the right
infrastructure in place to support them."
The time to bridge this gap is now, before the
chasm grows too wide to safely cross. Look at the challenge of
aligning business and IT as an opportunity to more clearly define
your organization's goals and to reach them more quickly than
you ever dreamed possible. T
Stephen Poole has written about computers and technology for a variety of national magazines
since 1988.
PHOTO BY ALEX HAYDEN
CHESS PIECES BY BOB PONZONI
|
Quick Links
