The launch of Jasmine, a pure object
database developed by Computer Associates alongwith Fujitsu, and its encouraging
international reviews may signal that the technology is now ripe for commercial markets.
Jasmine is a combination of object database, Internet development platform, and multimedia
authoring tool. A team of 1200 engineers from Fujitsu, and 200 from CA were on the design
board for roughly two-and-a-half years to get Jasmine out. Even before that, a Fujitsu
team had been working at object database development. But CA won”t be the first company to
stride down the object database path. Oracle and Informix decided to go the hybrid way,
while Sybase took an all-object database strategy. None have yet succeeded by any measure.
In light of this, CA”s move comes under close watch.
CA”s partner in India for the information
This will be followed by distribution of
The question is-is it time yet for object
databases? "If object-oriented development and need for image-laden content is any
measure, then yes," says Martin Rennhacckamp, Contributing Editor, DBMS magazine, who
was a keynote speaker at the database conference in Mumbai. However, if one considers the
fact that many of the organizations worldover, may be particularly in India, are still
grappling with the aftermath of their early- to mid-nineties deployment of applications on
client-server RDBMSs, it is too early to consider an architectural shift in their database
strategy. Therefore, the vendors thought of taking a hybrid approach-object relational
databases (ORDBMS). Oracle”s 8.x Universal Server, Informix Universal Server, and IBM”s
Universal Database all sought to unify the entire issue between relational and object
technologies, while CA took the stand that hybrid approach won”t work. Explained John
Venema during his recent visit to India, "In fact, in December”95, we too had
announced our plans for OpenIngres/ODBMS, our own universal server. But within three
months we realized that the hybrid approach won”t work." Yogesh Gupta, Senior VP
(Product Strategy), CA, who was part of Sanjay Kumar”s entourage, reasoned out that such
approaches never worked in the past also. For instance, relational capabilities could not
be extended to hierarchical and network databases of the mainframe era. Therefore, CA now
Jasmine”s pure object architecture
eliminates the extensive overhead associated with ”relational”izing objects. Together with
SQL class libraries it can reach into existing data repositories, thus extending the full
benefits of object technology without sacrificing the investments in relational data and
applications. Therefore, it is more suited for dynamic multimedia applications over
corporate networks or the Internet while delivering the integrity and practical data
management required for mainstream business solutions. Coupled with this, is a code-free
multimedia development environment.
One of the critical success factors is
Jasmine”s developer movement. Product architectures have a direct impact on vendor”s
channel strategy. An architecture that is more open, possibly with published APIs, and
portable across multiple platforms, sets the stage for other potential partners to expand
the overall market for the database vendor”s product line.
Does this spell the end of the ORDBMS era?
No. There are two factors that go in favor of ORDBMS. First, they are nothing but
extensions to RDBMSs, a proven product category. "RDBMSs are mature and are more
finetuned for optimized performance and provide a very rich set of functionality,
including support of advanced features like parallel processing, replication, high
availability, security, and distribution," says Rennhacckamp. Second, the ORDBMS
vendors have immense marketing muscle and can wield developer power. The more risk averse
users will continue to try out new applications on RDBMSs. But, despite the presence of
other object database vendors like O2, Versant, and Object Design, CA”s announcement
assumes significance because of its reach into the corporate world.
Even CA is quite modest about its market
expectations. Said Venema, "We are not naive to think that we are going to capture
the database market. It”s really a five-year project." So right now, CA is whipping
up developer and user interest in this category of products and educating them. Nor is it
"adding a baggage for OpenIngres to carry around," said Venema, because the
installed base of OpenIngres is not the primary target.
An IDC document, authored by Steve McClure,
sums up the issue nicely: "Putting object extensions on RDBMSs is tantamount to
adding stereo radios on horse-drawn carriages. You will have interesting enhancements but
the wrong base vehicle."