1993, a vendor of artificial intelligence software named Trinzic acquired Channel
Computing of Portsmouth, NH and inherited a product called InfoPump. InfoPump
was a script-based data movement tool for the portion of the data warehousing
market known as Extract/Transform/Load (ETL) tools, and was a market leader
in the early 1990's. In 1995, Platinum technology International, inc. purchased
Trinzic. The large influx of research & development capital from Platinum allowed
the InfoPump developers to greatly enhance the 3.0 version of the product. Unfortunately,
at the same time, market analysts began to predict the demise of scripted data-movement
tools. The belief was that graphical tools were necessary to reduce the need
for programmers and increase use by the business analysts who actually owned
the data. In 1997 Platinum began development of DecisionBase 1.0, a combination
of a GUI mapping tool (developed in-house), Platinum's Open Edition Repository
version 1.6 (for metadata management, technology acquired by Platinum's purchase
of the Reltech and Brownstone companies), and InfoPump 3.2 (for data movement).
1.0 was released in March of 1998, but GUI mapping functionality was severely
limited in the initial release. For instance, the mapper always assumed a row
that was being moved to a target database was an insert and the code had to
be manually modified to allow an update. By March of 1999, with release 1.9,
a significant amount of new functionality had been added, including the ability
to bulk-load data to Oracle and IBM UDB, support for Microsoft SQL Server 7.0,
the ability to do pre- and post-processing, and the ability to modify the generated
SQL from within the mapper.
Magazine favorably reviewed the product in March of 1999. The major drawback
was the DecisionBase price tag, which was listed at approximately $200,000,
depending on the number of metadata scanners and database interfaces necessary.
This was the major contributor in the size of the installed base, which at last
count was approximately 40 customers.
market for ETL tools is expected to grow from $327 million dollars in 1996 to
$620 million dollars by 2001, an increase of almost 90%. The number of vendors
in the ETL market in the mid-1990's was small, comprised of basically four companies
(Prism, Carleton, Evolutionary Technologies, Trinzic) plus some modest offerings
from IBM. In the past four years, the space has become very crowded, with over
fifty vendors competing in various market niches (e.g. specializing in access
to VSAM databases). Four vendors still primarily control the general market,
including Ardent, Computer Associates, Informatica, and Sagent, with some offerings
from IBM and Oracle. Prism has merged with Ardent, Carleton with Apertus, and
Platinum with Computer Associates. The major vendors are now working furiously
to find a competitive differentiator, with the most popular differentiator being
integration with Enterprise Resource Planning packages such as SAP and PeopleSoft.
Strategy and Trajectory
Associates' DecisionBase strategy is to create a new product, to be known as
"DecisionBase TND" (The Next Dimension). This product will include:
Integration with CA Unicenter TNG (The Next Generation)
to provide integrated workflow services and replace Platinum's "Synergy"
product. Unicenter will also provide high-speed data transport services,
probably via CA's "TransportIT" product (probability 80%).
Integration with CA Jasmine TND Object Oriented Database.
This relates to CA's "Aspen" repository effort, which is the new Microsoft
Repository that Platinum co-developed with Microsoft.
Integration with Platinum's Forest & Trees product for
data visualization. We believe portions of this integration are a result
of an ongoing internal Platinum effort known as "Project C" (an integrated
decision support infrastructure suite which would provide services that
unify multi-vendor decision support tools)
technology's main competitive advantage is the metadata management aspect of
the product, as it is one of only two true enterprise metadata repositories
on the market. For customers interested in a true fully functional repository
capable of functions like impact analysis and data rationalization, this is
a strong selling point. However, for customers only interested in data movement,
the price tag associated with the inclusion of the repository is a difficult
sell. The other advantage is that the underlying data movement engine is the
InfoPump product, which is feature-rich and has a very large installed base.
Computer Associates appears to be positioning the product as a strategic end-to-end
data warehouse solution. As stated in their press release: "A Data Warehousing
strategy must be based on the most powerful technology for collecting all the
information that might possibly be relevant, regardless of its form and its
location - in-house, the Web or other public data - through a powerful infrastructure,
and through intelligent data collection tools. And to collect, understand, to
correlate and leverage this information, through comprehensive metadata management.
And to analyze it, using intelligence technologies. And to present the results
effectively, through sophisticated visualization technology. And to deliver
it, through modern user interfaces and the Web." It is likely that this strategy
will be refined over the next few years as CA acquires additional software companies
and their technologies.
The product exhibits strength in its metadata management
layer, especially for companies with on-going metadata management projects.
Platinum's repository technology was industry-leading, as evidenced both
by its large installed base and the fact that it was chosen by Microsoft
to co-develop the repository included in Microsoft SQL Server 7.0 (see http://msdn.microsoft.com/repository/).
The repository technology is a significant competitive advantage with customers
interested in metadata, since none of CA's closest competitors have a competitive
Data Movement Engine: the InfoPump data movement technology
is well known in the industry for its robustness, is very flexible and has
a strong development language. It has the capability to handle virtually
any datatype, including exact numeric and support for Binary Large Objects
Database interfaces: Another key differentiator is the
fact that the interfaces which interact with the source and target databases
are written to the native C API ("Application Programming Interface" for
the C/C++ programming languages) supplied by the vendor, where possible.
Many other vendors use generic ODBC (Open DataBase Connectivity, a Microsoft
technology that is becoming a de-facto standard) for connectivity. Using
the C API (for example Microsoft and Sybase call theirs "DBLib" and "CTLib",
Oracle's is "OCI") allows the interface to take advantage of specific vendor
features, some of which are quite powerful. ODBC is the least common denominator,
so these specific features become unavailable. The C API also allows for
support of non-standard datatypes that the vendor may implement, while the
ODBC standard only supports a limited number of datatypes.
Integration with CA Unicenter will provide CA with access
to its large installed base. Depending on the pricing model CA chooses,
customers may regard it as just another Unicenter "plug-in".
CA will probably improve the method currently used to access
legacy "flat files". A more intuitive approach to mainframe and UNIX file
systems would be a big improvement.
Computer Associates only recently completed the acquisition
of Platinum technology. Since CA previously had virtually no presence in
the data warehousing space, and a great number of Platinum employees either
were not retained by CA or left on their own, it will likely be some time
before the CA sales force can effectively market the DecisionBase product.
Another consequence of CA's new arrival to data warehousing
will be market perception. Customers about to set out on expensive, long-term
data warehouse projects will want strong assurances that CA is committed
to data warehousing, both from a research and development perspective, and
from a product line perspective. Customers who have doubt about CA's commitment
to the product will go elsewhere.
Platinum was tardy in porting DecisionBase to the most
popular UNIX platforms (HP/UX, Sun Solaris, and IBM AIX), while DecisionBase's
closest competitors are already strong on those platforms. CA's position
on UNIX ports has not been publicly announced, and DecisionBase is only
available currently on Sun Solaris, and only in beta.
CA also lags behind the competition in Enterprise Resource
Planning (ERP) integration. The product does have the ability to read SAP
data, but there is no SAP BW integration (already present in the Ardent
product), and no integration with PeopleSoft (present in both Ardent DataStage
and Informatica PowerMart). If industry rumors are correct and CA acquires
PeopleSoft, integration may occur sooner, but as with all CA acquisitions,
integration of the corporate cultures and products would be long and painful.
Release dates have been pushed back and many customers
are frustrated. Since the CA acquisition of Platinum, a number of key employees
have left the company or had their positions redefined. This, in conjunction
with the fact that the product future has been completely changed, will
make it difficult for CA to meet the announced "Q4 1999" beta release date.
Cost. DecisionBase costs a great deal more than many of
the competing technologies (from 50% to as much as 500%).
Widgets and Wizards. Competing products have pre-defined
"widgets" (code snippets to perform more complex processing) which can be
dragged and dropped into the GUI. DecisionBase is currently woefully short
in this area, and provides no pre-defined widgets with the shipping product.
DecisionBase is also short in the area of wizards, assistants that walk
the user through a "question and answer" session to help fill out complex
Mainframe data access. Some competing products have embedded
technology to access non-DB2 data residing on IBM MVS mainframes, as well
as other systems, often without the need to convert the data into a "flat
file". Ardent's acquisition of Prism should give them a leg up on CA in
InfoPump upgrade path. There is no way to migrate existing
InfoPump code into the repository. This means that existing InfoPump customers
can not upgrade to DecisionBase unless they are willing to re-write all
of the work they have already done.
CA should make sure that enough developers are committed
to the product to ensure timely releases.
DecisionBase is already behind competing products in its
feature set, so any further delay will hurt sales.
A strong effort should also be made to complete the UNIX
ports in a timely manner.
As many pre-defined widgets as possible should be included.
Mainframe data access needs to be improved.
CA should increase the amount of pure product development
effort to add needed features, especially in relation to the effort being
expended on integration with other CA products. DecisionBase is a viable
product, but it lives in a highly competitive market, which is growing rapidly.
The product needs to be managed with this in mind.
contains powerful technology which customers should investigate. If there is
a strong need for metadata management, DecisionBase with its Open Edition Repository
is a good candidate. Current CA Unicenter customers may also find the forthcoming
integration with DecisionBase a powerful feature. DecisionBase's use of native
database interfaces will also be important to some users, especially ones who
use the more precise datatypes available in some RDBMS's (i.e. Oracle's Number
datatype when used with a greater precision/scale than the standard Float datatype,
the Exact Numeric Decimal datatypes in Sybase and Microsoft).
to pricing, DecisionBase is not a good candidate for customers who have no interest
in metadata management, or for smaller data mart implementations. Incomplete
integration with ERP products also makes DecisionBase a poor candidate for customers
who wish to extract data from SAP or PeopleSoft.
its incomplete integration with mainframe data access products, DecisionBase
may also not be the appropriate product for customers with a heavy reliance
on legacy mainframe data sources (i.e. IMS, IDMS, and VSAM).