| Craig S. Mullins
The Database Wars
By Craig S. Mullins
As I write this in late November 1998, the smoke is just beginning to clear from Comdex and one thing is clear: the database wars are heating up all over again. Just when you thought it was all over. Just when you thought the DBMS was becoming a commodity. Just when things were threatening to get boring.
The First Punch
Microsoft threw the first punch at Comdex, but it was a telegraphed punch all the way. Everybody knew that Microsoft would be announcing SQL Server 7.0 at this huge conference in Las Vegas, Nevada. And Microsoft got a lot of positive coverage for the SQL Server 7.0 launch.
Truthfully, there are quite a few things to like about this new version. The addition of row-level locking enables the major ERP vendors to support Microsoft SQL Server as a database platform. With Oracle concentrating more and more on its applications business, Microsoft SQL Server will look better and better to major ERP vendors like SAP, PeopleSoft, and Baan. Another new and improved feature is the way memory is sized; SQL Server 7.0 uses dynamic memory allocation to adapt memory usage based on application requirements and the amount of memory available for use. Microsoft has also improved the speed of many processes, its space management abilities, and claims to provide better overall scalability.
In addition to the new and improved internal features of SQL Server, Microsoft is bundling more functionality into the base product. An online analytical processing engine called OLAP Services is integrated with SQL Server 7.0. This raises the bar for business intelligence by providing some of the key reporting and analysis functionality right out-of-the-box with the database engine. Additionally, Microsoft is bundling a data movement and transformation tool called Data Transformation Services and the Microsoft Repository 2.0 for metadata management.
Furthermore, Microsoft seems to be receiving strong early support from customers and vendors. Microsoft claims to have more than 10 major corporations that already have deployed SQL Server 7.0 in production. And Microsoft suggests that by the end of 1999 there will be more than 3,000 SQL Server 7.0-optimized applications available.
For all of its effort, Microsoft SQL Server 7.0 won the Best of Show Award at Comdex. Sounds great, right? But the one thing that was generally ignored is that the product is not generally available and will not ship until early 1999. During the product launch keynote at Comdex Microsoft President Steve Ballmer indicated that SQL Server 7.0 would be available in early January; even though Microsoft had previously said the product would ship by the end of 1998. Missed deadlines are not uncommon in the software industry, but usually more folks complain about it and the media is more critical than we are seeing in this case.
So, as you read this in January, the product might have shipped, but then again, it might not have. Isn't it kind of funny to have a product that isn't shipping win awards and generate the amount of media frenzy that SQL Server 7.0 has generated?
Oracle did not just sit back and concede the "database buzz" to Microsoft. In fact, Oracle parried with the old one-two. The first punch was a $1 million challenge and the second was a sucker punch to the belly of Microsoft's operating system business.
Larry Ellison, CEO of Oracle Corporation, delivered a keynote speech at Comdex during which he offered $1 million to anyone who can run a specified query in SQL Server 7.0 that achieves better than one-one-hundredth the speed of the same query against an Oracle database. On November 23, 1998, Oracle clarified its challenge in a press release that specified the following:
"Oracle Corporation will pay $1,000,000 to the first person who can demonstrate that SQL Server 7.0, with a 1-terabyte TPC-D database, can come within 100 times of Oracle's best published performance for query number 5 of the current TPC-D specification (Version 1.3.1)." Additional details of Oracle's challenge to SQL Server 7.0 can be found on Oracle's Web site at http://www.oracle.com/challenge."
This is an interesting challenge and one that anyone will be hard-pressed to win. Oracle is running their benchmarks on highly scalable Unix machines, whereas SQL Server 7.0 can only run on Windows NT, and the new, more scalable version of Windows NT is not due from Microsoft until later in 1999.
Ellison then followed with an attack on Microsoft's operating system business when he unveiled Raw Iron. This project will enable Oracle8i to run on a "server appliance" without an operating system. The idea is that by eliminating the overhead of the operating system, database performance can be improved.
Raw Iron is an audacious announcement. It has a huge potential upside if Oracle can remove the requirement for an operating system. However, the hardware (and the software) has yet to be produced. This could be just another NC, but I doubt it. If Oracle can convince major hardware suppliers like HP, IBM, Sun, and Compaq to support Raw Iron, why would a database server require an operating system? It has a chance to succeed, but it requires several things. First, users will need an interface to a file system for creating database files, but Oracle8i provides that with its IFS (Internet File System). Furthermore, it needs a robust enterprise management environment including functionality for performance management, database administration, database utility execution (load, reorganization, unload, etc.), security implementation, etc.
The Bottom Line
Even though the DBMS vendors are hurtling along with new features and enhanced functionality, the user adoption rate of new releases is slow. Users have not migrated to Oracle8 from Oracle7.x as rapidly as Oracle would have hoped. With the Year 2000 crisis in full bloom in 1999, many organizations will be fully consumed with priorities greater than moving to a new DBMS release even if performance, functionality, and scalability gains can be achieved.
On the plus side, the intensifying database wars between Oracle and Microsoft (and IBM) will help to drive innovation and reduce initial cost of implementation. This should benefit organizations and database administrators. However, with innovation often comes increased complexity, which in turn increases management costs and increases the overall TCO of database development.
As far as who will win the wars, my guess is no one, at least for the immediate future. For heterogeneous environments, Oracle is currently the clear choice. But don't discount IBM, yet. IBM plans a mid-1999 splash for DB2 Version 6, the latest and greatest release of DB2 Universal Database Server.
The Windows NT market is even more hotly contested. According to the industry analyst firm Dataquest, Oracle outsold Microsoft on Windows NT platforms in 1997. It is reasonable to assume that this spending pattern held steady in 1998 because nothing new happened to change customer-purchasing trends. But with all of this hoopla and still more to come, 1999 is definitely up for grabs. Stay tuned throughout the year for more details on the ongoing database wars!
From Computing News&Review, January 1999.
Business email through exchange hosting services provided online.