Sybase, DBMS Clusters, MPP, and DATAllegro

Sybase is a DBMS stalwart that gets far less attention than deserved. The company recently beat financial-performance estimates and has raised its 2008 sales estimate to $1.11 billion. Sybase's on-going success — the company's DBMS is much more than the parent of Microsoft SQL Server — earns the company a closer look, in its own right and as an excuse for one last comment on Microsoft's planned acquisition of DATAllegro.Start with the 2nd quarter financial results. According to CEO John S. Chen, "our strong 38% growth in database license revenue... was driven by mainly robust demand for both our ASE flagship database, all the options associated with it and our IQ analytics server. In the quarter we attracted a total of 234 new ASE customers as well as 48 new IQ customers."

ASE is Adaptive Server Enterprise, targeted to high-availability, high-volume transaction processing. The company launched a Cluster Edition last February. It uses a shared-disk architecture (which contrasts with the shared-nothing architecture of leading MPP systems) to extend DBMS scalability and reliability.

IQ is Sybase's analytics-optimized server. IQ does not support non-clustered parallelized processing, unlike column-store + MPP rival Vertica. Nonetheless, IQ is the column-store market leader: the company claims more than 1,200 customers with more than 2,000 implementations. On the strength of IQ, Gartner places Sybase just barely outside the leader's quadrant in the September 2007 Magic Quadrant for Data Warehouse Database Management Systems. The May 2008 launch of the Sybase Analytic Appliance, packaging IQ with Microstrategy BI software on IBM hardware, should help the company enhance its position in the DW market.

Interestingly, Sybase tried MPP (massively parallel processing), the claim to fame for DATAllegro as well as Greenplum, Netezza, and others. Sybase's shared-nothing MPP launched in 1996 (as did IQ) as an option for Sybase SQL Server 11, but the company abandoned MPP in 2003.

Microsoft SQL Server was originally Sybase SQL Server running on the Windows platform. Version 6.0, released in 1995, was the first independent release, and the DBMS code was rewritten for 1998's version 7.0. Since the mid-1990s, Microsoft has released new versions of SQL Server every 2-3 years. And that's why, after reading DATAllegro's FAQ on its acquisition by Microsoft —

— I wrote that it would be 2-3 years before "a reliable, robust, parallelized, appliance-packaged SQL Server" would be available.

DATAllegro CEO Stuart Frost told me in a subsequent phone conversation, "I can see where the confusion comes from," that is, from the fact that Microsoft's and DATAllegro's published materials on the acquisition talk only about integration of DATAllegro's parallelization technology into SQL Server. The FAQ does not say that pending the parallelization of SQL Server, Microsoft will ship a DATAllegro appliance with the simple substitutions of SQL Server for Ingres and Windows for Linux, which Stuart says is actually planned. In fact, the sentence "DATAllegro's products will not be sold to new customers until they are fully integrated with SQL Server" — the direction as written is not SQL Server into DATAllegro's products — says the opposite. So attribute the misunderstanding to miscommunication.

Stuart explained, "our technology provides a wrapper that turns pretty much any standard DBMS into shared-nothing MPP. During our feasibility tests, we found SQL Server to be ideal for our approach — to the point where it will actually be much easier to integrate than Ingres." And he elaborated, "DATAllegro's software will become a separate SKU for Microsoft."

Stuart now understands that my writing that Microsoft would be "gutting" DATAllegro referred to the product and not to the company. Microsoft will move forward only with DATAllegro's propriety software components and not with the current appliance's Ingres DBMS or Linux operating system. (I'm not clear on the intended hardware direction.) Regarding the company, Stuart says, "one of the main reasons why MS is keeping us down here in California is to maintain a high level of autonomy and to ensure we can maintain as much 'startup agility' as possible." That's a good thing.

I wish DATAllegro the best of luck as a Microsoft component. Change is good; the continued fast pace of the evolution of information technology, and of business models, benefits users and solutions providers alike. The Sybase and DATAllegro examples both prove that adaptability is key to on-going success.

Added Thursday, July 31, 2008, 9:30 PM

I note that the "About DATAllegro" description in all the PR that has gone out on the deal carries the pre-announcement text "DATAllegro v3 goes beyond the high performance of first generation data warehouse appliances and adds the flexibility and scalability that only an open platform can offer."

With the move to SQL Server and Windows, DATAllegro will be leaving behind the platform's open software components. Each operating system and DBMS has its advantages, but what's with continuing to tout advantages that are being abandoned?

Don't these people read the material they put out?Sybase is a DBMS stalwart that gets far less attention than deserved. The company recently beat financial-performance estimates and has raised its 2008 sales estimate to $1.11 billion. Sybase's on-going success — the company's DBMS is much more than the parent of Microsoft SQL Server — earns the company a closer look, in its own right and as an excuse for one last comment on Microsoft's planned acquisition of DATAllegro.