Columns
The rise of analytic packages
- By Wayne W. Eckerson
- December 31, 2002
There has been a great deal of talk about and interest in packaged analytic
applications recently. There is a race among vendors to duplicate SAP's success
in operational applications and become the ''SAP of analytic applications.''
Business Objects, Informatica and Hyperion, as well as ERP giants like SAP,
PeopleSoft and Oracle, have delivered domain-specific analytic packages that
provide 65% to 85% of a finished application.
Although analytics packages may be the talk of the town
today, most organizations still build analytic applications by a 2-1 margin,
according to a recent TDWI report titled
The rise of analytic applications: Build or buy?
(To download the full report, go to www.tdwi.org.)
The burden of maintaining code
Today, most organizations that
build analytic applications write a lot of code to surround an analytical tool,
such as Business Objects WebIntelligence. In fact, half of the organizations
surveyed by TDWI reported that in-house code represents 50% or more of their
custom analytic applications. The hand-written code typically provides the
organization's look and feel, integrates the tool into a corporate portal
environment, or adds missing functionality.
It is a huge burden for organizations to develop and maintain custom code
over time. As a result, many organizations are eager to ditch in-house
development and migrate to packages. Not surprisingly, the percentage of
organizations that are ''interested'' or ''very interested'' in purchasing packaged
analytic applications will increase from 47% to 62% in 18 months. In addition,
the ''negative rating'' of analytic packages will plummet from its current 43% to
19% in 18 months. Clearly, many companies would like to avoid writing code if at
all possible.
Accelerating time to insight
Although packages are gaining
momentum, there is a new breed of development tools that give organizations a
way to accelerate the deployment of custom-built analytic applications and make
them easy to upgrade. Called analytic development platforms, or ADPs, these
tools offer ''plug-and-develop'' capabilities that enable developers -- or even
savvy business users -- to build sophisticated applications with a unique look,
feel and functionality in a matter of days or weeks.
AlphaBlox, arcplan, Proclarity and Business Objects are some of the companies
delivering first-generation ADPs today.
''I've never seen anything out-of-the-box that gives us what we want,'' said
Rick Stevenson, director of financial systems and supply chain at GAF Materials
Corp. in Wayne, N.J. Stevenson uses arcplan's Dynasight to develop and maintain
an analytic application for the sales department. ''Our [analytic development
platform] is very easy to use and requires no coding,'' he said. ''I had a request
for a significant enhancement yesterday and I built the new functionality in one
hour.''
ADPs are a new breed of analytical tools that expose their object models in
the form of graphical, configurable components, such as chart, grid, sort,
filter, query and access control components. Developers can drag and drop the
user interface components onto a graphical workbench that facilitates WYSIWYG
development. The client and server components can be easily customized using
pop-up dialog boxes and options for creating scripts, if needed.
Most ADPs run against multidimensional databases, such as those from
Microsoft, Hyperion and Oracle, but a growing number also work directly against
relational data in a data warehouse or data mart. For example, Business Objects
now sells Application Foundation for building custom analytic applications. And
MicroStrategy is migrating its toolset to support a greater degree of customized
development.
The best ADPs support sophisticated query engines that can dynamically
display from disparate systems as distinct objects (such as charts, tables or
graphs) in a single report on the user's screen. In addition, many analytic
development platforms come with pre-defined models and reports for building
applications in specific domains that developers can leverage to accelerate
development.
Re-creating the EIS of old
In many respects, the emerging
analytic development platforms are re-creating the ''executive information
system'' (EIS) of the late 1980s, which provided compound graphical reports,
including charts, tables and documents, for top executives. The difference is
that those early EISs ran on mainframes and were hand-coded by a team of
developers dedicated to supporting the executive office.
The main difference between EISs and ADPs is that ADPs run on standard
operating environments and don't require a team of developers to support and
maintain them. In fact, ADPs take custom coding out of the process, greatly
accelerating deployment times, or ''time to insight.'' In addition, ADPs make it
possible for organizations to rapidly deploy EIS-like applications to all users,
not just executives.
Build and buy
ADPs fulfill a need that exists in most
organizations to deliver robust functionality in a customized environment.
Packages also make this promise, but many organizations so far have failed to
gain the benefits of packaged analytic applications because they spend too much
time customizing the packages. This leaves an opening for ADPs to fulfill the
majority of organizations' analytic needs.
More than likely, however, organizations will pursue both a build and buy
approach. And vendors will meet these needs. Packages will contain ADPs to
rapidly customize the program to meet user needs and navigate the shoals of
version upgrades. ADPs will come with more robust domain-specific models and
report templates to accelerate custom development.
About the Author
Wayne W. Eckerson is director of education and research for The Data Warehousing Institute, where he oversees TDWI's educational curriculum, member publications, and various research and consulting services. He has published and spoken extensively on data warehousing and business intelligence subjects since 1994.