The first quarter of 2003 is now behind us and it was an interesting one, indeed. The DBMS market was rocked by news both good and bad from all of the major vendors. We had worms attacking our SQL
Servers, IBM announcing a major new version, and Oracle earning profits not seen since the early days of the century. And that is just a quick surface view.
First up this quarter is a new research note from IDC Corporation with the ominous title “Oracle Hears Footsteps.” The basic premise of the research note is that even though Oracle still leads
the DBMS market, its biggest competitors – IBM and Mircosoft – are gaining rapidly.
According to the report, Oracle’s market share slipped to 39.4% last year from the 41.7% share it held in 2001. This translates to a 5% drop in share for the leader. The same report shows IBM
recording a 9% gain to reach 33.6% of the market – which is less than 6% behind Oracle. At the same time Microsoft is really gaining steam with a huge 15% increase. Although the overall Microsoft
market share is still tiny compared to Oracle and IBM, it is looking healthier than ever at 11.1% of the market.
Even more interesting are some of the insights IDC claims to be behind these trends. The report indicates that Oracle depends on large customers with increasing capacity needs and that the economic
slowdown has caused many of these customers to defer their capacity upgrades. Additionally, the report claims that there are Oracle customers with excess capacity that has yet to be deployed,
further eroding Oracle’s revenue. IBM, on the other hand, is more stable with large customers due to the stable entrenched base of mainframe and AS/400 customers. The report then couples this
stable environment with growth on Linux, Unix, and Windows platforms, as well as growing CRM implementations on DB2. Finally, Microsoft garners success by focusing on the small to medium sized
business market.
All in all, the IDC report neatly summarizes the current DBMS market landscape. But what really happened this past quarter? Let’s dig into the details.
A Worm Named Slammer
In late January 2003 Microsoft SQL Servers were attacked by an Internet worm that took advantage of a well-known vulnerability in SQL Server 2000. The Slammer worm, as it is called, uses a buffer
overflow in SQL Server to take over the system and send out a flood of packets. This flaw has been known since Summer 2002 and Microsoft had a patch available already when the attacks began.
Microsoft continues to urge its customers to patch their SQL Servers. More information about the patch can be found at … http://www.microsoft.com/technet/treeview/default.asp?url=/technet/security/bulletin/MS02-039.asp.
The worm caused significant Internet slowdowns with the Asia Pacific region being hit particularly hard. At the height of its impact, the Slammer worm was reported to have performed about 1 billion
scans an hour as it looked for new Internet hosts to infect. Keynote Systems, Inc., which maintains an index of the major US Internet sites, reported an average slowdown of approximately 50% while
the Slammer worm was running amuck. The overall impact of the Slammer worm was a significant slowing of Internet traffic.
The intriguing thing about this worm is that it did not have to cause any problems at all. Microsoft had supplied patches months earlier and, if vigilant administrators had applied them, there
would have been no SQL Server vulnerability to exploit. Of course, this is an unfair analysis of the situation, I suppose. Should it be more incumbent upon the administrators to rapidly apply fixes
or for the DBMS vendor to supply a reliable DBMS to begin with? Hmmm…
At any rate, Slammer may indeed be the sign of a trend that will haunt DBAs for years to come. Successful worms and viruses usually result in additional, similar worms and viruses. So our databases
will likely be under attack again in the near future. And next time, the hackers might exploit an unknown vulnerability. For more information about worms and Slammer check out a great article on
preventing SQL Server worms at http://www.dbazine.com/cook7.html.
The Quarterly News From Oracle
Early in the quarter, the first week to be exact, Jeff Henley, Oracle’s CFO and a member of its board of directors, announced that Oracle might consider paying a dividend on their earnings if the
government eliminates dividend taxes. The announcement came right after President Bush proposed getting rid of dividend taxes as part of a ten year, $674 billion economic stimulus package. The U.S.
Congress, of course, must first approve the package. And that appears unlikely to happen, at least in its original form, with the current Iraqi war going on.
If dividend taxes are eliminated I would not be surprised if Oracle – and many other technology companies – begin to pay dividends. Speaking of earnings, Oracle’s third quarter fiscal year 2003
results came in and there was some pretty good news – but all was not completely rosy. Profit grew 12 percent – which was significant as this marks Oracle’s first year over year profit improvement
since November 2001. The company earned $571.3 million (11 cents a share) in the three months ended February 28, 2003. In the same period a year ago earnings were $508 million (9 cents a share).
And, in more good news, revenue was up almost 3 percent to $2.31 billion from $2.25 billion a year ago. And revenue from Oracle’s software license updates and product support jumped 16 percent, to
$1 billion.
So what’s not to like? Well, new software license sales were down. And deferred revenue was down sequentially to $1225.2 million in fiscal Q3 from $1248.3 million in the prior quarter and up from
$1013.8 million in fiscal Q3 2002. And for the news that is near and dear to our hearts, new database software license sales were $602.9 million, down 4 percent from the same period last year.
Revenue from new sales of applications was off 5 percent, at $140 million.
But what looms in Oracle’s immediate future? In their quarterly earnings conference call, Oracle CFO Jeff Henley predicted fourth-quarter earnings ranging from 12 cents to 15 cents a share. Henley
called rising oil prices and anxiety about the war in Iraq the “wild card,” and said that “capital decisions tend to suffer when there are rising levels of uncertainty.” Analyst estimates for
Oracle’s fourth quarter are 14 cents per share.
Larry Ellison also took the opportunity to slam Microsoft. He said “A lot of people say, well, Microsoft is eating us alive at the low end in terms of gaining market share at the low end because
Microsoft costs less. Well, it’s actually very expensive when your web site and your whole business goes down because of worms. When the SQL slammer worm basically took down all the Microsoft
databases, that was very expensive…the Oracle database stayed up, the Microsoft database stayed down.”
But let’s put that in perspective. InfoWorld columnist Robert X. Cringely reported in March that Oracle uses Microsoft SQL Server databases. Cringely reports “That’s right, and chief Larry
Ellison was forced to suck it up and admit this, albeit internally, when Oracle was struck by the SQL Slammer virus.” Cringely goes on to cite an internal memo written by Ellison confirming that
the Slammer worm slowed Oracle’s networks. In the same column he also reported that the Oracle Development Tools User Group uses Microsoft’s Jet database engine because “retaining an
Oracle-based shop for the management of the service would be prohibitively expensive.” So maybe Microsoft really is gaining share at the “low end,” huh?
And Over at Big Blue
Each quarter we have analyzed Oracle’s corporate performance, but not any of the other big DBMS vendors. Well, here we will reverse that trend by taking a look at IBM’s results. On January 16,
2003, IBM reported fourth quarter fiscal year 2002 results beating consensus earnings per share and revenue estimates. But IBM requires a bit more poking around to get the information we are
interested in. Let’s face it folks, IBM is a huge corporation. Revenue in the quarter alone was $23.7 billion – yes that was billion with a “B.”
First of all, keep in mind that software is IBM’s third largest business – behind hardware (which is first) and services (which is second). IBM had better than expected results from its services
unit which drove the revenue growth for the quarter. Software contributed $3.8 billion in revenue which represents 16% of total revenue. This is actually a 2% decline over the same period last
year. IBM’s data management software was down 3% for the quarter, but grew 9% for the full year. When DB2 is broken out, though, it shows growth of 11% for the quarter and 9% for the full year.
So, at least the DB2 software business is healthy at IBM.
Which brings us to this quarter’s technology news from IBM. First of all, on February 4, 2003, IBM unleashed its information integration strategy and beta products. The first two products from its
federated data integration project are DB2 Information Integrator and DB2 Information Integrator for Content. Both products enable customers to abstract a common data model across data and content
sources and to access and manipulate them as though they were a single source. This is the beginning of the delivery of IBM’s federated database management approach. The key with federation is to
enable businesses to access and integrate data from multiple locations as if it were stored in a single location. This is a big announcement and will offer users of the technology the ability to
integrate data from multiple sources, both relational and nonrelational. Look for IBM to continue highlighting this strategy and these products over the course of the next few quarters. More
details can be found at http://www.ibmlink.ibm.com/usalets&parms=H_203-021.
Secondly, in early February, IBM introduced DB2 Universal Database Express Edition, aka DB2 Express. IBM states that DB2 Express is specially tailored for the Small to Medium sized business (SMB)
market (companies with 100 to 1,000 employees). DB2 Express is very obviously targeted at Microsoft and its success at getting SQL Server installed into smaller accounts. Highlighted by features
such as lower pricing, simpler installation, pre-configuration for specific target businesses, and support for Linux and Windows, IBM will look for greater low- to mid-market success with DB2
Express.
Next up, in late February IBM announced a beta program for DB2 .NET Enablement. The highlight of the program is the new DB2 .NET Data Provider which is written in managed C# code to deliver
high-performing, secure access to DB2 data. This shows that IBM is supporting Microsoft’s .NET initiative and will be providing methods for DB2 to work within its guidelines. To learn more about
IBM .NET Enable beta program visit: http://www7b.software.ibm.com/dmdd/downloads/dotnetbeta/
And finally (yes, it was a busy quarter for them), IBM announces the details of Version 8 of DB2 for z/OS (that’s the mainframe one). Although V8 has been out for several months now on Linux,
Unix, and Windows platforms, it has just been announced for mainframe platforms. There is a reason for this – DB2 for z/OS Version 8 is a huge and complex beast. Let’s take a look at some of the
major components of this new version.
One of the biggest impacts of V8 will be the requirement to run it on a zSeries machine with z/OS v1.3. That is, DB2 V8 will not support old hardware nor will it support OS/390 (IBM’s mainframe
operating system prior to z/OS). Additionally, DB2 customers must migrate to V7 before converting to V8. There will be no IBM-supported capability to jump from V6 (or an older version) directly to
V8 like there was in V7. This is significant because fewer than half of IBM’s mainframe DB2 customers have migrated to V7.
DB2 V8 will have the ability to support large virtual memory. So, theoretically, with 64-bit addressing DB2 could have up to 16 exabytes of virtual storage addressability to be used by a single DB2
address space.
Broader usage of Unicode is another architectural highlight of DB2 V8. V7 delivered support for Unicode-encoded data, but V8 forces its use. If you do not use Unicode today, you will when you move
to DB2 V8 because the system catalog will be implemented using Unicode.
Long DB2 object names is another significant architectural change in V8. DB2 V8 significantly increases the maximum length of most DB2 object names. For example, instead of being limited to 18 byte
table names, you will be able to use up to 128 bytes to name your DB2 tables; the same limit applies to most DB2 objects and special registers including views, aliases, indexes, collections,
schemas, triggers, and distinct types. The limit for columns is 30 bytes, but table spaces and packages are still 8 bytes, unless it is a trigger package, which can be 128 bytes. This brings a lot
of flexibility, but also a lot of reworking of the DB2 catalog tables.
One such reworking requires the use of table spaces with 8K, 16K, and 32K page sizes. Therefore, the system catalog in DB2 V8 will require use of the BP8K0, BP16K0, and BP32K buffer pools.
DB2 V8 offers many enhancements to DB2 objects including sequence objects, variable length index keys, expanded partitions, new types of partitioned indexes, new partition management, and
materialized query tables (also known as automated summary tables). Also, index keys can comprise up to 2000 bytes – so more data can be indexed using a single index.
DB2 V8 also introduces Schema Evolution. Today, there are many types of DB2 changes that require the DBA to DROP and then re-CREATE the object in order to implement the change. Schema evolution
enables the DBA to make more types of changes to database objects using native DB2 features.
Other administration highlights include support for up to 4096 partitions per partitioned table space, row-level security, session variables (for global security), and Data Partitioned Secondary
Indexes (DPSI). DPSIs are significant because they resolve the issue of dealing with non-partitioned indexes (NPIs) on tables in a partitioned table space.
DB2 V8 also offers row-level security, so applications that need a more granular security can be supported. Finally, it looks like we will be able to have partitioning and clustering be independent
of one another. In other words, the clustering index key can be different than the partitioning index key.
Numerous SQL and programming features are being added to DB2 V8. For example, SQL improvements include the ability to get diagnostic information, true SEQUENCEs, dynamic scrollable cursors,
multiple DISTINCT clauses, qualified column names on the SET clause of INSERT and UPDATE statements, the ability to mix EBCDIC, ASCII, and Unicode columns in a single SQL statement, and the ability
to SELECT from an INSERT statement. In addition, V8 offers significant changes to the SQL system limits.
For Java programmers DB2 V8 offers expanded functionality in the form of support for both Type 2 and Type 4 Java drivers. Both will be updated to support the JDBC/SQLJ 3.0 standard which brings
enhanced support for things like SAVEPOINTs and WITH HOLD cursors, as well as improvements to connection pooling, and a long list of other expanded features. And finally, in DB2 V8 more XML support
is being pushed into the DB2 engine.
And, as with every previous new DB2 version, IBM is making significant enhancements to improve application performance. DB2 V8 optimization enhancements will include sophisticated query rewrite
capabilities to support materialized query tables, sparse indexing to improve star join performance, support for parallel sort, and better support for queries with data type and length mismatches
which would have caused less efficient access paths in previous releases.
Truly, DB2 V8 is the most significant new version of DB2 ever in terms of new functionality and the amount of new engine code that IBM is writing. The announcement letter for DB2 for z/OS Version 8
can be found at http://isource.ibm.com/cgi-bin/goto?it=usa_annred&on=203-019.
Other Quarterly News
In the middle of January 2003 MySQL, the Swedish company that sells the open-source database of the same name, announced a release of MySQL that runs on IBM mainframes. The MySQL DBMS for the
mainframe will run under Linux on IBM’s zSeries mainframes. This version of MySQL is available for download, and was showcased at the LinuxWorld Conference in New York. Although I do not
anticipate a mad rush of customers running MySQL on mainframes, it does highlight a couple of market trends, namely:
- An increase in the popularity of open source software
- The continued health of the mainframe
- Growing support for Linux on IBM hardware
Summary
And so ends the first quarter of 2003. All of the major DBMS vendors were making news this past quarter and it looks like the remainder of the year will be just as interesting and turbulent as
ever. So be sure to check back with TDAN.com next quarter to hear all the latest news about the DBMS market place.
