Analysis: Behind Sun's MySQL Acquisition
Will MySQL transform Sun into a BI and DW force to be reckoned with?
Sun Microsystems Inc. last week gobbled up open source relational database specialist MySQL AB in a $1 billion move that some industry watchers are calling an out-and-out bargain.
Until now, Sun has been mostly a peripheral player in the business intelligence (BI) software and services market segment.
It has several long-standing partnerships (most notably with both Sybase Inc. and Oracle Corp. in the large-footprint data warehousing arena) as well as an innovative partnership with data warehouse appliance specialist Greenplum Inc., but it's shown nothing like the commitment of rivals Hewlett-Packard Co. and IBM Corp. to BI or even data warehousing. The MySQL acquisition, which effectively hands Sun a competitive and very popular RDBMS environment, alters that calculus considerably. Whether it will transform Sun into a BI and DW force to be reckoned with is, experts say, the $64,000 question.
Sun CEO Jonathan Schwartz didn't talk up a potential BI or data warehousing renaissance in post-acquisition commentary on his blog last week. Instead, he sought to pump up The Big Picture.
"[N]o platform vendor has assembled all the core elements of a completely open source operating system for the Internet," Schwartz wrote. "With this acquisition, we will have done just that -- positioned Sun at the center of the Web, as the definitive provider of high performance platforms for the Web economy."
MySQL is an integral BI and DW player in a couple of important respects. First, it has partnerships with several big name BI players, including the former Business Objects, which bundles MySQL as the default repository with its BI tools. MySQL is also a significant player in the open source BI installations (along with PostGreSQL), and has a partnership with Pentaho, developer of an all-open source BI suite.
Philip Russom, senior manager of research with The Data Warehousing Institute (TDWI), sees the MySQL acquisition as of a piece with an ongoing open source acquisition frenzy.
"It would seem that open source vendors are on the block, given recent acquisitions of JBoss [by Red Hat], Sleepycat [by Oracle], ZenSource [by Citrix], and Zimbra [by Yahoo]," he points out. "Good for Sun: they grabbed while the grabbing is good."
On the other hand, Russom doesn't know if Sun can complete the transition from a hardware-focused Unix OEM to enterprise software provider. "The commitment is there and initial investments are made, but that's no guarantee. We could say the same for HP, right now, and I recall that HP tried this transformation and failed back in the mid 1990s," he argues.
"Clearly, hardware is [a commodity], so the profitability -- and speed to new products -- of software looks appealing. IBM saw this coming in the 1990s and transformed its DB2 and WebSphere product lines into mega platforms that dominate applications development and information management."
Russom, at any rate, doesn't see MySQL as recasting Sun as a relational database powerhouse -- at least not overnight.
"I wouldn't say that open source databases have any where near the popularity -- or capability -- of Linux. My experience tells me that open source databases are categorically 'swiss cheese' compared to real, enterprise databases, like the leading relational database management systems," he says.
"To be fair, the holes are natural, since one of the assumptions of open source is that the user will fill the holes themselves, to achieve certain unique requirements. But I've seen users overlook the development work required, then end up late and over budget with projects involving open source databases."
The Popularity of PostGres
Nor does Russom believe MySQL will spur Sun's data warehousing ambitions. Another open source RDBMS -- PostGreSQL -- would have been a better choice in this regard. "MySQL is rare in data warehousing, whereas PostGres is by far the open source database I find most," he indicates. "Even so, I know a lot of people who build turnkey operational applications, and they prefer MySQL. I guess that an opinion I've heard multiple times from users is true: MySQL is best for transaction processing, but PostGres has better data loading and query capabilities."
Veteran data warehousing architect and author Mark Madsen agrees.
"MySQL is still relatively low on the list of BI databases. If you include the embedded use in appliances and the like, Postgres and Ingres probably have a greater number of individual users than MySQL when it comes to strict BI usage. I do think [the acquisition] will help MySQL get further traction in mainstream IT," Madsen comments.
"The interesting thing is that MySQL largely owns the database market when it comes to Web applications. Startups, new technology, Internet businesses are all running MySQL because it's designed for low-cost scaling across many nodes. It doesn't try to go head-to-head with Oracle or DB2 with large, single-instance databases, so I think the impact to database vendors is relatively low. The real impact on those vendors is that MySQL has carved out that (relatively) new and growing part of the market for itself."
MySQL isn't the sole provider of its special sauce, so to speak. When used for high-volume transactional applications -- i.e., the very applications in which it's carved out a niche for itself -- it typically uses its InnoDB storage engine. InnoDB software is available under the GNU General Public License, but it's currently developed by Oracle, which acquired InnoDB via its acquisition of the former Innobase Oy in late 2005.
"Also, when you look at MySQL there's one thing that can be confusing: engines. Basic stuff is deployed with the ISAM engine. If you poll all the huge Web sites out there -- e.g. Google runs YouTube on MySQL -- you'll find that they use InnoDB," Madsen explains. "When you need transactions or performance, it's InnoDB. Then there's SolidDB, provided by Solid, which IBM bought not too long ago. If you look at it, MySQL really needs a decent storage engine to perform specific tasks -- great architecture, but if you're an appliance guy, you need the full set, not a framework sans engine."
Madsen doesn't think it wise to put MySQL in the same class -- for all applications -- with its bigger, more established, proprietary RDBMS brethren.
"You need to watch the query response question [with MySQL,] too, because I think it depends on the type of query, complexity, etc. I've seen some tests withMySQL table scan rates that [took twice as long as] Oracle's," he points out. "The issue with all these guys is a lot less experience with single-node and fine-grained parallelism, concurrent locking, cache management, and the hundred other things that make a database like Oracle run as well as it does."
There's also the question of just how Sun plans to market MySQL. If Oracle and DB2 are effectively out of reach, Microsoft -- whose SQL Server 2005 database is essentially in the same class, performance-wise, as both Oracle and DB2 -- isn't. That's because SQL Server 2005 runs only on Windows, whereas both Oracle and DB2 run on a variety of platforms -- including Linux.
Even though Sun and Microsoft have cozied up to one another to an almost unbelievable degree (both vendors claim to have put their traditional competitiveness behind them), industry watchers don't completely buy it. Sun and Microsoft won't necessarily be contesting just the database itself. MySQL is yoked to the open source Linux, Apache, MySQL, and PHP/Perl/Python (LAMP) stack, which -- as a platform used to support dynamic Web sites -- competes with Microsoft's .NET architecture, too. It's for this reason, argues TDWI's Russom, that Sun will probably set its sights on Microsoft.
"We all know about Sun's history of competitiveness with Microsoft, so from a Sun mentality viewpoint, acquiring MySQL makes sense because it helps Sun's commitment to LAMP compete more strongly against Microsoft's .NET," he says.
Mike Schiff, a principal with data warehousing consultancy MAS Strategies and a contributing analyst for data management with Current Analysis, agrees.
"While major database vendors including IBM and Oracle are not likely to find themselves in heated competition with MySQL -- other than in situations where cost is a critical concern and many advanced database features are not required -- Microsoft, based on its history of battles with Sun … is likely to be more directly impacted by the acquisition," he points out.
"Although MySQL can run on Windows and Linux as well as Solaris, Microsoft SQL Server can only run on Windows. Furthermore, many consider Microsoft's .NET and LAMP to be competitive technology stacks. With the resources of a major company such as Sun behind it, MySQL becomes more competitive with Microsoft technology."