- Don't look back
- Is support for OSS optional in your business?
- Nokia N810 Tablet + WiMax
- Vendors need to right-size their products
- Dolphins Invade Sun Campus!
- State of Open Source
- MySQL Workbench: open source data modeling
- Comments on The 451 Group's Database Report & Red Hat's 4Q revenue
- Kaplan: Guiding open source in IT
- Can the transportation market teach us anything about the software market?
October 10, 2005 | Comments: (0)
Why InnoDB matters so much to MySQL (really)
I don't think the open source industry fully appreciates the importance of Oracle's purchase last week of Innobase and what that means for MySQL.
MySQL management (see customer email) was immediately on record as saying that the GPL protects MySQL because it guarantees that the InnoDB table manager remains available for anyone, including Oracle or MySQL, to enhance and extend. That's right. They also say that this doesn't matter, and that MySQL's customers and business are unaffected. That's not right.
During 2003 and 2004, MySQL relied on a dual licensing strategy (the same as Sleepycat and other *embedded* open source vendors use for their products). That was a risky thing to do, because as a standalone server, they have considerably less leverage over their adopters. Most people don't have to write code to use MySQL, so there's less reason for their customers to pay.
Dual licensing works only for so long as you own, or have rights, to all the IP that you dual license. That's why MySQL signed a separate agreement with Innobase in the first place -- it gave MySQL the legal standing to dual-license the MySQL plus InnoDB combination. They still have an agreement in place (for less than a year to go) for that, but if they want to renew it. Do you think the price may go up? Will the sun rise in the East?
MySQL could make the argument that they need no separate agreement. The MySQL server itself is released under the GPL, so may link and use the InnoDB engine. No MySQL customer changes the database server under any circumstances, so customers are insulated from the terms of the combination.
In fact, if you look carefully at the licensing arrangement that MySQL's paid licensees enter into, you discover that it's not the database server that forces them to buy a commercial license. It's the client library that knows how to talk to the server. There's no Innobase code in that client library, so the dual licensing business can continue.
It's not quite that simple, though. The lower the perceived value of the software that customers are paying for, the lower the price they'll be willing to pay. If customers are aware that they're really just licensing a client library, they're going to want to pay ODBC driver pricing -- single dollars per copy -- instead of database server pricing (hundreds/thousands of dollars). MySQL skirts that issue now by talking little about the actual restrictions that the GPL imposes. There is, likewise, a real risk that the open source community or a proprietary vendor will produce a free competing client library, and take away even the small lever that MySQL holds today.
For MySQL in the long run, you don't want a business model that depends on your customers being dumb and frightened. That is not Marten Mickos' style anyways.
Marten realizes that and has been working hard to move from dual licensing to subscriptions for his primary revenue source (like Red Hat did). This makes sense. Most MySQL deployments are in IT shops where no software gets distributed; people build apps for internal use. They don't need to release the code they write, because the GPL requires that only if they ship binaries. Since they don't, they can use the database server under the GPL, and have no reason to pay for a different license.
The problem is that the VAR and OEM licensing business that MySQL built up over the preceding years is still the biggest contributor to the top line (they just inked a big new deal with HP in August, for example). The OEM deals are big, perhaps even million-dollar range, and thus material to MySQL's march towards a liquidity event. Subscription adoption is growing, but if the OEM licensing revenue stream were to dry up, rumors of a 2006 IPO would change the last digit to an 8.
Besides that, the MySQL sales force -- really, any sales force -- isn't going to master the complexities of the viral GPL requirements in the clients and how the Inno acquisition affects grants of rights to the server. They're not going to want to tell all their existing paid licensees that they really didn't have to pay hundreds per copy after all. The sales team will be confused when customers ask hard questions, and with the Oracle guys in the field, you can be sure that customers will have a printout of hard questions to ask.
You have to assume that Marten was never able to convince Heikki Tuuri (Innobase CEO) to sell to him, but that's because he was offering MySQL stock and little cash. Oracle isn't cash-constrained; it could afford to pay cash money in this deal.
Smart money in the industry would bet that the acquisition price was less than $5M, and probably between $2M and $3M. Oracle is a smart buyer and doesn't overpay as a rule. They'd view this expense as a marketing investment if they had to. This is spare change and for Ellison a modest investment in freezing the market. Confusing a competitor's cusotmers is easy to get approval for from Oracle management. Can you say PeopleSoft?
This isn't deadly to MySQL. They have a great product and a good business. The timing is really breathtaking -- exactly when the 5.0 release comes out with the Oracle features they've lacked for so long, Oracle confuses everyone. Marten has attracted Oracle's attention for real, now, and is forced to respond to an agenda other than his own. We've seen this movie before. It doesn't take a genius to predict that Oracle's marketing and sales machine now kicks into gear and begins to deposition MySQL more aggressively.
Let's hope Marten and team are prepared. And they may want to look at swapping out InnoDB with Berkeley DB as the table manager in their product.
Posted by Dave Rosenberg on October 10, 2005 10:28 AM
RATE THIS ARTICLE:
-
- COMMENTS
Hmm, you didn't mention MySQL Network (http://www.mysql.com/network/) at all, the subscription offering for end users.
With MySQL Network, customers gain access a package of value important to them, licensing doesn't come into it though users have the option of a non-GPL license (at no additional cost) if their company prefers this.
Our deal with HP is also about MySQL Network.
Regards, Arjen Lentz
Community Relations Manager, MySQL AB
I don't buy it. If the MySQL sales force does not master the complexities of that oh so scary and complex GPL, it's because their customers don't care. If MySQL customers don't care, then they will pay less and be happy. If MySQL customers do care about licensing restrictions, then a quick comparison of the GPL (you can do anything at all, except place restrictions on sub-licensees) with the restrictions Oracle has in their licenses should make it pretty plain where the advantage lies.
Of course, some few may want to pay (more?) for a database they can use to lock-in their licensees, and some of these may want to pay someone besides MySQL for the privilege. So what? Nobody _cares_ whether MySQL makes money hand-over-fist. All anybody cares about is getting a good database at a good price, and this is what you get when you pay MySQL for support. The "Open Source Industry" _knows_ that they are all about eliminating economic rents by commoditizing software. It is all about _shrinking_ overall license revenue and making money the old fashioned way, by saving the customer time or effort -- not by charging customers for non-rival goods. As the idea of Open Source sinks into the business community you're going to see happier customers paying less. This is not a problem, this is a goal. Anybody who's expecting the economics of Open Source to be otherwise is deluded.
Posted by: Karl O. Pinc at October 10, 2005 08:51 PMI really like this part:
-- cut --
And they may want to look at swapping out InnoDB with Berkeley DB as the table manager in their product.
-- cut --
Have you looked close to this task ? How much work is it ?
Answer: one line in one file.
MySQL AB does not joke about "pluggable back-ends". Berkeley DB is already supported - just not in default configuration. So if InnoDB will become unavailable the switch will take few days (to recompile all binaries and do some testing): code is already there.
Of course Oracle can buy SleepyCat as well...
Posted by: Khim at October 10, 2005 10:37 PMOf course, this moves does have the affect of giving MySQL the ability to market their product as being "built on Oracle". I am not a marketing specialist but there must be some benefit to being tied through code to such a big and technically respected player.
Posted by: Scott Alan Miller at October 11, 2005 05:18 AMQuestion for y'all:
How do you think this will affect postgres?
I think the significance of the Oracle/InnoDB acquisition goes way deeper than what may appear.
Oracle just threw MySQLs customers into a huge pile of uncertainty by buying the transaction engine that implements MySQL 5.0 advanced database features, which was announced only weeks ago in late September. What does MySQL have left without InnoBase? Their Classic edition is what, with a flat-file DB engine that certainly does not support triggers or stored procedures or transactional capabilities.
These customers are also left to wonder what the community support around MySQL is as now clearly appears -- that all most critical development and everyday support is strictly contained within a few individuals. MySQL is more about business than community. And they made such a business blunder as to rely so heavily on external product that is now on the hands of Oracle. Now, what happens when Oracle decides to put a little squeeze on that license cost or heavily influence the evolution and support of that product? MySQL customers and prospects will have to reevaluate their database investments and analyze short and long term impact on their businesses.
People just need to recognize the Oracle InnoBase acquisition for what it is, and MySQL customers need to have their wits about them over the next several months as this deal plays out.
Posted by: Paola Lubet at October 11, 2005 02:25 PMWell... I see the following as worst case scenario
for mysql:
1- mysql is distributed with no InnoDB support
2- mysql is distributed with InnoDB and is expensive
Case 1:
Really, 99% of mysql users don't care about it.
Mysql is strong on web development business.
Anyone who requires high transactional capability
is using something else. Most distros do not even
enable InnoDB support in the default mysql configuration
Case 2:
99% of the users will stay with mysql, since
they do not need transaction support. The remaining
1% will think twice to keep with mysql. Some will
use postgres or firebird or sapdb/maxdb or oracle or even sql server. MySQL keeps 99.8% of its customers.
Postgres will keep ascending now that they natively
support win32.
I don't think Oracle can or aimed at replacing MySQL position in Internet/Web app environment. It wants to prevent MySQL entering Enterprise market (where transcation is needed)
Posted by: goken at October 11, 2005 07:16 PMJust use Ingres.
http://opensource.ca.com/projects/ingres/
Posted by: Ganesh Prasad at October 11, 2005 08:39 PM" with a flat-file DB engine that certainly does not support triggers or stored procedures or transactional capabilities"
Wrong,wrong,right.
But then there is BerkeleyDB for that last one.
Read the documentation.
http://dev.mysql.com/doc/refman/5.0/en/triggers.html
http://dev.mysql.com/doc/refman/5.0/en/stored-procedures.html
http://dev.mysql.com/doc/refman/5.0/en/bdb-storage-engine.html
Heh. And now Oracle has acquired Sleepycat Software. Oracle now owns both of MySQL's best back end data stores. What interesting times we live in.
Posted by: Art at February 14, 2006 07:53 PMhttp://dev.mysql.com/doc/refman/5.0/en/triggers.html
http://dev.mysql.com/doc/refman/5.0/en/triggers.html
Wrong,wrong,right.
But then there is BerkeleyDB for that last one.http://dev.mysql.com/doc/refman/5.0/en/triggers.html
http://dev.mysql.com/doc/refman/5.0/en/triggers.html
Wrong,wrong,right.
But then there is BerkeleyDB for that last one.http://dev.mysql.com/doc/refman/5.0/en/triggers.html
http://dev.mysql.com/doc/refman/5.0/en/triggers.html
Wrong,wrong,right.
But then there is BerkeleyDB for that last one.
- Get Started
- Port 25 Blogs
- OSS News
- Join a Project
{Open Source} Heroes Happen Here
Start today and order your own Hero Hack Pack – which includes Getting Started with Open Source, Windows Server 2008 and Visual Studio 2008 Trial. Each pack is a chance to win a free pass to OSCON 2008.
TOP STORIES
ADDITIONAL RESOURCES
- Remote Access: Maintain Security and Decrease the Burden on IT
- Beyond AntiVirus: Symantec Endpoint Protection
- What Every Enterprise Needs to Know About VDI
- Disaster Recovery in Minutes
- Protecting Microsoft(R) Applications
- Reduce Recovery Times and Tape Costs
