One of the things business has learned over the last 2-3 decades is that diversification is not always to best way to go. A lot of companies have given up on diversification to get back to their core competencies.
Although Oracle appears to be 6 times larger than Informix, when you compare just the engine (license) sales, the difference is only a little more than 2x. Informix is showing growth rates in engine license sales in the 50-70% range, considerably than Oracle's 30-40%. If this rate of growth continues, Informix can close that gap rather quickly.
As for the OO situation, there are two keys to remember. First is time to market. Recent stories have indicated that "cartridges will not be in the June release of Oracle8 (assuming that it meets that date, it will only be 3 years late). Therefore, it will be surprising to see cartridges before the end of 1997. On the other hand, developers have had actual platforms to develop DataBlades on for more than one year (because the Illustra engine has been shipping that long). There are a lot of similarities between Illustra and the initial version of US, so most DataBlades should move relatively easily. For those who have been developing to the DataBlade Developers Toolkit that was shipped in May, they had both specs and a working platform for testing (i.e., Illustra could run them for testing purposes). What can Oracle say in this area? That is why Informix can claim a 2 year lead.
The second reason is performance. Much of Oracle's solutions are bolt together of single purpose servers or other pieces. Since each piece knows little or nothing about the others, there can be little optimization--hence poor performance. On the other hand, DataBlades can be integrated into the server and the server has full knowledge of the requirements and capabilities. This allows query optimization on the basis of cost, creation of indices on new data type contents, and optomization of storage (between normal tables vs. BLOB's), as just a few of the advantages. Oracle with its one server to store text data, and another to store spatial data, and no way to create custom data types is way behind. |