Scott
Leibs, CFO Magazine
February 01,
2003
Cognos Inc.'s successful bid for rival Adaytum
Inc. last December (see "Adaytum: An Addendum," below) is just one
indication that the business-intelligence (BI) software market is
heating up. Analysts expect further consolidation as software firms
battle in a large ($11 billion, according to IDC Corp.) but
slow-growing market.
While the companies that make core BI software fight it out, a
host of developments in related areas also promise to improve
customers' ability to crunch through vast stores of data and make
smarter business decisions. One reason BI and data warehousing can
be so costly is that companies are continually adding to the stores
of information they need to process, thus driving up total storage
costs even as cost-per-unit declines. One answer may be application
data management (ADM), a method for enforcing data-growth and
-retention policies by moving unused data off of production
databases while still providing access to it through whatever
applications require it. This software traffic cop differs from
traditional archiving methods because it is application-specific and
works across distributed environments. Consulting group Gartner
predicts a 64 percent compound annual growth rate for ADM software
through 2006.
While ADM allows companies to make the most of the hardware
they've got, a case can also be made for new gear. Consider Netezza
Corp., a highly touted start-up that recently rolled out its first
tera-scale data "appliance," a combination of hardware, software,
and storage designed specifically to tackle BI. Combining servers,
storage, and database into a single unit, the Netezza Performance
Server line is billed as providing 10 to 20 times the speed of
conventional processing at half the cost.
The Netezza server doesn't replace BI software, but sits
underneath it to provide a horsepower boost. Steve Duplessie,
founder of consulting firm Enterprise Storage Group, says the
product is a breakthrough, with the potential to radically change
how companies approach analytics. By shifting BI processing to a
special-purpose system, companies can avoid overloading current
infrastructure with what is often a very demanding form of
processing. They can also, Netezza argues, spend far less time
preparing data to be analyzed and more time actually analyzing it.
Netezza's 8000 line of appliances is priced from $622,000.
Netezza, in effect, provides a ready-to-use data warehouse, which
may appeal to companies that have resisted the large investments
that traditional data warehouses require. Kalido Ltd., a relatively
new entrant in the data warehouse space (it was spun off from Royal
Dutch/Shell Group of Cos. two years ago), will soon offer new
options. The company's current product, Dynamic Information
Warehouse (DIW), will be offered in a more modular format beginning
next month, with customers having the option of buying only
interfaces (to specific ERP systems, for example) they need.
More notable is a new Kalido product dubbed "reference data
management." Reference or "master" data is the term for information
that changes, as opposed to transactional information, which
doesn't. The two most common forms are customer and product data.
Both can bedevil analytics because they may be referred to
differently in different parts of a company, and often reside across
a number of systems. The Kalido reference data management engine can
peer into all those systems and produce a "golden copy" that is
always the most up-to-date and reliable.,/p>
Kalido chief technology officer Cliff Longman says that "one
problem with data warehouses is that they are often built for one
purpose, and prove inflexible as needs change." Kalido, which thus
far has won most of its business among large multinationals has yet
to determine pricing for its reference data management engine. Its
DIW product is priced from $300,000.
Application Data Management at a Glance
- Utility: Enhanced response times when working with active data
- Potential Savings: $125,000 per terabyte after first terabyte
- Optimum Usage Scenario: Large data sets and high number of
mirrored data instances
- Avoid: On new data warehouse/data mart implementations and/or
small database footprints
