Briefing Note Sun MicroSystems and Sybase have a sufficiently strong partnership that they go around doing joint briefings, yet it hardly gets much publicity, writes Phil Howard. Why is this? I had better explain what their partnership entails.

Around 12 months ago, Sun published a reference architecture for using its hardware in conjunction with Sybase IQM (IQ Multiplex). While the former is well known the latter will be unfamiliar to many readers. The reason for this is twofold.

In the first instance, Sybase IQM is a database designed specifically to support data warehousing. It is a relational database but it stores and retrieves data by column instead of by row. This has some significant advantages, notably that the resulting database is very much smaller than is the case with conventional relational databases, and that it is very much faster (by orders of magnitude) at retrieving data, particularly when complex queries are involved. A third big advantage is that Rcubes (relational cubes) provided support for much simpler database schemas than is possible using traditional approaches, and implementation is therefore much faster and easier.

So, if it's so good - why haven't you heard more about it? Well, one reason is that Sybase decided a few years ago that it would focus the product on the VLDB (very large database - typically measured in tens of terabytes) market, where the main players are IBM and Teradata. Since this is a relatively small sector of the market, the product's marketing was similarly constrained.

However, it was this emphasis on VLDB that interested Sun, because Sun has not, historically, had much penetration in the high-end data warehousing sector. So a partnership between the two companies seemed an obvious move. As I have said, this resulted in the publication of a reference architecture.

What this means is that Sun tested all sorts of different hardware/software scenarios, investigating the best options to ensure availability, the use and positioning of servers, storage options, network choices and so on, resulting in recommendations that provide optimal performance for particular environments. Further, you can go into an iForce Solution Centre and test your own data against the reference architecture as a proof of concept.

So far, so good. But this would remain only of minority interest if it was restricted to the VLDB space. However, about six months ago Sun persuaded Sybase that there was no reason why it shouldn't target customers lower down the market - the only real difference was that you didn't need as much of the multiplexing that provides IQM with its scalability - all the other advantages of the product would apply equally well at any level within the market. So, Sun and Sybase are now addressing what they call the EDM (economical data management) market which, to you and I, more or less means anyone in the data warehousing market that isn't a VLDB.

In my view, Sybase IQM (which will, of course, run on platforms other than Sun) offers some significant advantages over the use of conventional relational databases.

However, it has one marketing problem: companies like PeopleSoft, with its Enterprise Warehouse, don't support Sybase IQM as the underlying database for its warehouse (because the necessary transformations would have to be re-written for column-by-column loading as opposed to row-based loading). And, basically, they won't do so until Sybase has a sufficient head of marketing steam behind its product, despite the fact that they are offering less than best practice for their customers.

But it is difficult to get this sort of momentum without the support of the big application vendors. Hopefully, this article will create a little more headway to get Sybase out of this Catch-22 position. © IT-Analysis.com

Phil Howard's biog