Are Open Source Databases Following in Linux' Footsteps?
In the future, he asserted, Microsoft would be all but irrelevant by the Linux trend. The crowd nodded and smiled at his screed. This, you could almost hear them thinking, was Larry at his best.
But then someone in the audience took the flipside to his thesis, asking Ellison: If Linux, in all of its open-source glory, would irreparably harm Microsoft's proprietary approach, could he not see the threat of open-source database vendors to his company? Ellison said no, and then enumerated Oracle's market strength, brand, track record and high level of security as reasons for Oracle's staying power.
Few can dispute those points. Oracle, which specializes in selling to high-end customers, is still a $10 billion company at the top of the database market. In the most recent calculations from research firm Gartner, Oracle retained a whopping 43 percent of the overall relational database market. IBM and Microsoft, which also cater to small and mid-sized businesses with their database products, are the other two most successful vendors in the vaunted space, with shares of 24 and 17 percent, respectively. It should be noted that IBM, which bumps heads with Oracle often in the quest for large customer contracts, has posted impressive gains in the database sector, and led during the last quarter in terms of new, licensed users.
But while commercial database vendors remain locked in a lunge and parry fight for customers, in the background is a small movement in which providers are open-sourcing their database as a less expensive alternative to the established vendors. This open-source movement, which some experts predict will follow a similar arc to Linux, is led by a Swedish firm that has been quietly building momentum as an alternative in the market for small business, departmental or commodity use: MySQL.
Industry opinion is divided as to whether MySQL or another of its ilk might challenge Oracle, Microsoft or IBM at the low end of the database market, where fewer specialty features are needed.
IDC analyst Carl Olofson summed up open-source databases technology: "Open source relational database management system (RDBMS) software is a disruptive technology. As such, it is used, not to address the high end requirements of those who demand the latest and greatest, but to address the needs of application and tools developers and vendors who are looking for just enough database technology to provide an affordable solution."
This is certainly comparable to Linux, which has enjoyed growth propelled by many major vendors, such as IBM and Oracle, as well as myriad Linux providers like Red Hat and SCO. These firms have pumped many dollars and resources into the movement to provide an alternative to proprietary systems such as Unix and Microsoft's Windows operating system. Isn't it imaginable, therefore, that open-source databases might follow the same path? After all, Linux started as a small movement, too.
MySQL's value proposition (contrary to Oracle, Microsoft or IBM) is to provide the code for its MySQL database for free, under the free software/open source GNU General Public License (GPL) or a non-GPL commercial license, which they make money from through service fees. And it's working.
In the spirit of "If you build it, they will come," MySQL has been building and users have been coming: MySQL has seen an estimated 4 million installations and over 30,000 downloads per day for its flagship product. The company counts Yahoo!, Lucent Technologies and Sony Pictures Digital Entertainment among its customers. At a time when investor dollars are scant, the firm this month landed $19.5 million in funding, led by Benchmark Capital, and appointed Benchmark general partner Kevin Harvey to its board.
MySQL versus the big three
Impressive credentials and support? Perhaps. But why go to an open-source database when Oracle 9i, Microsoft SQL Server and IBM DB2 are held in such high regard? In short, for the same reason why businesses are moving to Linux platforms: to save on total-cost-of-ownership, among other areas. The importance of this cannot be understated in this weak economy. Gone are the days when CEOs were looking to pad the top line; here are the days where the CEOS scrutinize how much cash they can shave off the bottom line.
So, consider a few options from Oracle, Microsoft and IBM. Oracle offers two main packages, an enterprise edition geared for high-end use at $40,000 per processor and a standard edition for $15,000 per processor, which is targeted for small- and medium-sized businesses (SMBs). Similarly, Microsoft has a standard edition of SQL server for $4,999 per processor, and an enterprise-class offering for $19,999 per processor. IBM's DB2 portfolio features an enterprise edition for $25,000 per processors, and a workgroup edition for $7,500 per processor (comparable to Oracle and Microsoft's standard editions).
By contrast, MySQL offers its database on a much lower scale and cost to the casual user. MySQL CEO Marten Mickos discussed his company's value proposition and licensing model in a recent interview.
"We are what we call a 'second wave open source' company, meaning we have a functioning business model that is in harmony with our free software principles," Mickos said. "Our dual licensing allows us to offer our software free of charge under the GNU General Public License (GPL) while at the same time selling the same product under a regular commercial license. We can do this because we own our software and have the freedom to license it any way we wish."
Olofson explained where open-source database technologies, such as those provided by MySQL, and two of its brethren, PostgreSQL and Firebird, fit into the market.
"Although the huge majority of MySQL usage today derives from free open source downloads, MySQL hopes to move from the geek community to the business community through the embedded database model, and they are counting on this channel as a key to future revenue growth," Olofson said.
Mickos, who said the company's goal is to "make superior database technology available and affordable to all," said commercial licenses account for nearly two-thirds of its revenues. The remainder of the company's earnings come from support and services. MySQL charges a flat fee of $440 per server and customers can have as many users and as many CPUs as they want.
"Put this in context of the pricing of other DBMSs [database management systems] and you can understand why we sell so much," Mickos said.
Now, given the price comparisons between MySQL and the other, strictly commercial vendors, why isn't MySQL grabbing up market share by the fistful? Does MySQL not threaten the three giants? No, and Mickos and his firm don't claim to. There are many, many reasons for this, some of which Ellison alluded to last April. MySQL is new, much smaller than the database behemoths, and does not offer much of the special functionality of large database vendors, something that they say the high-end, more specialized customers require.
"We focus on the commoditized part of the market - the one in which performance, reliability, convenience and price are the determining factors," Mickos said. As such, Mickos said he and his outfit feel they are complementary to say, Oracle, or IBM, who offer some highly specialized features, as opposed to competitive. In practical terms, while Oracle may power the high-performance needs of a search firm, certain departments might use MySQL as a simple alternative.
What the commercial vendors say
It's tough to find members of the commercial side of things that would argue with Mickos. Jeff Jones, director of strategy data management at IBM, praised MySQL and its brethren, noted that they are commanded by smart people with some great ideas. But he still sees the pros of the commercial side.
"I think you can argue there is a focus and a clarity of requirement to the commercial side," Jones said. "IBM has been working with databases since the '70s and things are done differently. There is a different degree of ability, of relational theory, for what we do. We have to bring scalability optimization, we have to understand how to extend relational theory to handle things like XML, to extend search technology, integrate more analytic capability, federate integrated business intelligence."
Interestingly, IBM has a small database offering that can slide into the low-end slot alongside MySQL, in the recently released DB2 Express. Geared for the low-end market, or a handful of seats, DB2 Express is priced at $499 per server, or $99 per user.
Ken Jacobs, vice president of product strategy at Oracle, shares Jones' feelings about the main differences between the open-source database players and commercial vendors of today. But, like Ellison, he suggested that it is Microsoft that has the most to lose from open-source, arguing that the evolution of open-source databases like MySQL could chomp away at Microsoft's low-end market share.
"Microsoft may be the real target or victim here," he said. "When people make decisions to purchase a database one of the deciding factors to go with open-source is that it is cheap." Admitting that Oracle and IBM's products are not as cheap because they target customers with more specialized needs, Jacobs said any cannibalization from low-end products is likely to be in Microsoft's camp. Jacobs said Oracle does not feel threatened by MySQL, and, like Mickos, said he sees it as a complementary product. "These DBs are interesting and can serve a niche. In fact, I'd argue that MySQL customers are also Oracle customers who have outgrown SQL Server."
Sheryl Tullis, product manager for SQL server at Microsoft, begs to differ. First, she disputed the notion that that SQL server's success is tied to SMBs, as many people suggest. "In the recent Gartner figures, SQL server grew 16.8 percent, Oracle declined 20.5. That shows customers are seeing value in SQL Server."
"We think [open source databases] are a different segment than who we sell to," Tullis said. We see MySQL as being primarily for hobbyist technologies, or departmental use. We see it as something a developer might use if they do not want to go through IT procurement."
Tullis said if Microsoft's customers do need a lightweight database, Microsoft offers MSDE, or Microsoft SQL Desktop Edition, and embeddable engine developers or ISVs can use as a datastore. And it's free. Moreover, she argued that Microsoft has actually learned, through MySQL, how to create a "database community that is interesting and passionate."
One of the knocks on open-source databases is they don't have some of the features large-scale servers do, and are therefore not attractive to customers. Meta Group analyst Charlie Garry has heard the arguments about MySQL and its brethren not having the special functionality or scalability of a DB2 or a 9i. He has a counter argument.
"There is a line of thinking that Oracle overshot the market by having too much feature functionality," Garry said. "One is XML databases. There is not a whole lot of people who are storing data in XML format."
"Now, there's an interesting feature I would guarantee, most people have never even heard of it or knew it was in 9i," Garry said. Then there are things like a partitioning option. More than 60 percent of enterprises are running on 4 processors or less and 50 percent of the resources are unused. A lot of people don't need partitioning, or 64-bit indexing that are offered by commercial vendors, so these are not necessarily things that people want to pay for."
"The beauty of open-source databases is that that we often see a cycle among commercial vendors, where they to continue to add feature functions to trump that other guy," Garry said. "Open-source databases are forcing users to think what is important to me, and think about right-sizing. When you're a tech leader like Oracle, you run into this rut where you have to drive revenues, listen to investors who tell you need these esoteric features."
Page 2: Analysts and open-source advocates weigh in One would be tempted to think Oracle is alone in its assessment that Microsoft may be threatened by a MySQL down the road. Not so. Gary Ebersole, president and COO of upstart ANTS Software, said he could see reasons for Microsoft having concern about this issue. Ebersole speaks as a competitor of Oracle and IBM, and ANTS technology looks to eliminates database-locking conflicts in high workload applications that need to be frequently updated, such as messaging platforms or stock trading. Ebersole argues IBM and Oracle can't handle the concurrency of data ANTS can, and sees that as a way for the small company to insert itself among the big vendors.
"I would contend that MySQL is following the same trajectory that Linux has followed," said Ebersole, who sold his firm Clustra Systems to Sun Microsystems last year. "It's a strong, viable technology. MySQL is a very mature, stable product, whose disciplines are just as strong as Linux. 80 percent of database requirement is pretty basic SQL. Oracle has every possible, product that an SMB departmental application doesn't need, such as data warehousing features. They probably don't see the loss because their low-end share was never that strong. Oracle doesn't see the business their not getting."
However, he said Microsoft's database share is so rooted in the SMB space (though he admitted the pending Yukon server is aimed at the mid-market), that Microsoft might have reason to be concerned.
IDC's Olofson described the competitive positioning. MySQL "is not really cutting into the business of Oracle (or other mainstream RDBMS vendors) at the moment, because it is addressing a market segment that they do not serve today. Most businesses like the comfort they get from the security of knowing that a company like Oracle or IBM or Microsoft is managing their RDBMS technology and providing regular upgrades and support. MySQL is getting there on the service side, however, and may soon start to eat into the low end. At that point, Microsoft has more to fear than Oracle or IBM, but the open source incursion into the market may not end there. Open source vendors, as they gain credibility, may slowly work their way up from the bottom, putting more pressure on Oracle and their competitors to provide more and more high end innovations. But these innovations confine them to a stratospheric level of use where managing tens of thousands of concurrent users, handling arcane security requirements, supporting complex data types and mixed workloads of data and content, and managing very large and complex data warehouses are the key requirements. Database workloads outside of these areas become increasingly available to open source exploitation over time."
But the problem with all of this competitive finger pointing, is that SQL Server is not exactly what keeps Microsoft up and running as a business. That software giant has plenty of other revenue streams to rest its head on. Even if a MySQL were to eat away at its market share, it isn't likely that Microsoft will be that affected.
But Meta Group's Garry said Microsoft continues to push the value of its stack and while they're price performance is close to the larger guys, they are not overpriced beyond what people are willing to pay for.
"Microsoft isn't so dependent on DB revenue that they couldn't cut bundle their standard edition in their operating system for free," Garry said. "It certainly helps boost the value proposition versus Linux. But, I don't see any appreciable impact of open source databases on the commercial vendors until the end of the decade."
So, then the question becomes less about how open-source can take market share from commercial vendors, and more about what purpose it can serve to customers.
What the open-source advocates say
Josh Berkus, who runs consulting firm Aglio Database Solutions, as well as an active participant and advocate of the PostGreSQL open-source database movement, believes this is a valid argument.
"The "SMB" market is, in my experience, the most rapid adopter of open-source technology. I run an independent consulting shop, and 80% of my clients are small-to-medium business. Their evaluation of technology is usually based on only these factors: 1) How much does it cost? 2) Does it have the essential features I need, or can it be affordably customized to have them? 3) What kind of maintenance does it require?"
Berkus said open-source software, and solutions based on it, have historically lead proprietary packaged vendor software in low cost, "hands-free" reliability, and customizability.
"Largely this is because of the economics of software development; it simply doesn't pay for any multi-million dollar development company to create software for tiny market sectors with specialized needs or for "mom & pop" operations. So this market section, which is quite balkanized but constitutes a large portion of the economy, is left to small ISVs to service. And small ISVs increasingly use Open Source software to hold *their* costs down. This will remain true whether we are talking about conventional desktop in-office software, or some future "Web Services" architecture."
As for PostGreSQL, Berkus likes its potential. "The PostgreSQL project has a much more ambitious goal," than competing, he said. "We want to be the best RDBMS period, Open Source or proprietary. We have already made some steps toward this goal by designing features which no proprietary database has, like our support of more than a half-dozen procedural languages and support for certain advanced parts of the SQL standard, among other things."
Forward looking, forward thinking
Are MySQL and its fellow open-source database crusaders a threat right now? No. But that doesn't mean that open-source can't nibble on commercial vendors' pie slices in the future and it would be wise to adapt to the evolving market, according to IDC's Olofson.
"For the next few years, we don't see open source databases as a threat to Oracle, but over time Oracle must morph into a software platform provider that offers a complete managed environment, in order to be picked apart from the bottom by this new disruptive force in the market. Oracle is in fact moving in this direction with its application server technology and emphasis on complete integration of fundamental IT services through its software. IBM and Microsoft are moving in a similar direction, each seeking to be seen as a software platform provider, not merely a database vendor."
A proper question one might ask, then, is what will the business computing market look like in the future. Olofson's wonders: "Will we still have the Fortune 1000 with large data centers and in-house IT staffs, and myriad medium and small businesses with limited, locally managed computers? Or, will the lower end of the market migrate to outsourcing services and integrated self-managing ("autonomic") computer systems, causing database technology to become, for them, integrated components of suites of computing services where the details are left up to professional computing service providers (online or packaged software)? What impact will all that have on database vendors, large and small?
Right now, the answers remain elusive. But check back in five to ten years.