Category: digital

The “Global Database Market”

Today the European Commission expressed their concern about Oracle’s acquisition of Sun, framed in the context of the database market.

What strikes me as particularly funny in the reportage on this story is many analysts’ claims that MySQL has a small market share. I’ll pick on Gartner here:

“MySQL has extremely small market share,” said Andy Butler, an analyst with IT research firm Gartner. “The E.C. is losing sight of the big picture and is bringing about a more anticompetitive situation by focusing on one product at the expense of the other moving parts.”

This quote is pulled from a Wall Street Journal article that posits, “Many analysts believe the commission’s focus on MySQL is misplaced. MySQL’s share of the $19 billion global database market is approximately 1.5%. It is dwarfed not only by Oracle’s database business, but by those of competitors such as International Business Machines Corp. (IBM) and Microsoft Corp. (MSFT).”

Gartner and the WSJ are looking at the US$19 billion global database marketplace. That’s an interesting approach, given that the vast majority of MySQL users don’t pay a cent for their software. So their usage does not factor into this US$19 billion market at all.

Now, an Oracle 11g license is not cheap. Every license Oracle sells is a major bite out of the US$19 billion global database market. Every user of the MySQL Community Edition doesn’t even make a nibble.

Forrester Research published a study in 2008 that showed that “MySQL has the highest adoption and growth. MySQL continues to have the largest mindshare in the open source database market and has the highest number of paying customers for product support: an estimated 16,000.”

Gartner themselves, Evans Data Corporation, and other industry groups have released similar findings; findings that MySQL and Sun proudly annouce. Indeed, look at the quote at the bottom of that page:

“The future of the database market will be the standardization on MySQL.” – Charlie Garry, Meta Group

Here’s a little informal, unscientific data. Look at Google Trends’ output for MySQL, Oracle 11g, Postgres, db2, and Microsoft SQL Server. It’s quite clear that Google is finding far, FAR more interest MySQL than for any other product among their searches. If MySQL’s market share is so low, why do Google search users search for that term so much more frequently (factor in the tens, if not hundreds, of times)?

With all due respect, I’d suggest Mr. Butler at Gartner read his own company’s reports, and when considering the “global database market,” do so without the blinders of capitalization. The world of business is far more nuanced than can be accurately expressed with a dollar sign in front of one’s research.

Monty Program At The CLS And OSCON

In July several Monty Program employees attended the Community Leadership Summit and O’Reilly’s Open Source Convention (OSCON), myself included. While there we spoke to developers, partners, users, and interested bystanders of a number of SQL servers: MariaDB, MySQL, Postgres, and others.

The week started with the Community Leadership Summit. First, a big thank you to Jono Bacon for organizing a fantastic unconference, and allowing us to participate! Monty Program considers it our privilege to have helped to sponsor the event, and we look forward to attending and sponsoring it again in the future.

Being an unconference, the schedule is determined by the attendees. We scheduled a session entitled “What’s going on with MySQL?” to answer community questions about the possible future of the MySQL community. The session was very well atttended, with DBAs, developers, Sun employees, and Monty Program execs all in attendance. Also present was Bruce Perens, creator of the Open Source Definition, and Larry Rosen, author of the Open Software License and Academic Free License. To us, this was illustrative of the great interest by many community members in seeing the path Oracle chooses to take with MySQL in the future. Topics included possible scenarios, disaster planning for worst case scenarios, and best practices given Oracle’s relative silence with regards to their plans for MySQL (a topic addressed by Monty in a blog post earlier in July).

A video snippet of the session is available at Wikimedia. Have a look! (Requires a browser plugin capable of Theora playback)

On Sunday night the Postgres team had a release party for Postgres 8.4. Congratulations to the entire Postgres team! It’s a fantastic SQL server, and 8.4 has an impressive list of improvements. Monty and I were invited to the party, and we had a great time. There was a t-shirt auction, and Monty bought the t-shirt off Josh Berkus’ back. Trust me, you haven’t lived until you have seen Josh bare-chested, and a semi-inebriated Finnish hacker having nasal sex with his t-shirt. Classic. Thanks to the Postgres team for a really fun evening!

Postgres vs MySQL? The flame-war is a myth. Grow up. 🙂

There was a concerted attempt to recover for Monday morning, and the beginning of OSCON. It’s the premier open source conference, and we were delighted to have been invited to speak. Our talk was on Thursday, and was again well attended.

We divided our talk into three parts. First, I gave an overview of what MariaDB is, how the project works, our philosophy, and our methods. Second, Monty talked about the work we have done thus far, what’s already been incorporated into our MySQL fork, and what lies ahead. Finally, we opened the floor to questions. Thankfully a break was scheduled for after our talk, as the question period ran well over the session time limit. O’Reilly staff were very understanding, and graciously allowed us to keep using our session room.

One thing that was made clear is that we have not been communicating enough about exactly the kind of work we are doing. This is a failure on my part. In my defense, I have been somewhat reticent about pestering our development team for status updates, as I’m really keen on getting a GA release of MariaDB out ASAIMS (“as soon as it makes sense”). Mark Callaghan, one of our favorite users, was especially curious about what features and bugfixes we have implemented. He was quite cheerful about it, and declined to slap me for not being more communicative (thanks Mark!).

Some of these questions were answered in Monty’s portion of our presentation, and our slides are available from O’Reilly’s OSCON site. Grab the PDF and find out what we’ve been doing, and how we have been doing it!

Next week is our company meeting in Majorca, and I will be spending time with the dev team compiling a definitive list of what you can expect to find in the MariaDB GA release that is tentatively scheduled for later this month. I’ll blog about that as soon as I have the information, and also ensure it is put on the Maria wiki. Remember, we won’t release code we ourselves wouldn’t use in production. Thus, if the release date slips a bit, understand it’s because we’re committed to providing busy DBAs with production servers the very, very best.

Thanks again to Jono, the Postgres team, Tim O’R and Alison Randal at O’Reilly, and everyone else that made our week so fun and productive! And finally, apologies for the tardiness of this post. I have had house guests ever since returning from OSCON. 🙂

Sun: “GPL FUD is not a standard business practice.”

A few weeks ago I wrote a post about an acquaintance that had been given some very strange legal advice from a Sun sales rep. It generated a lot of interest, from both people inside Sun and those outside. Here’s what I have learned since:

1). Many people think that keeping your sources’ identities private makes the resulting information lies or FUD. They should talk to Woodward and Bernstein.

2). Many people cannot interpret a question mark at the end of a sentence. I was not making an indictment with my previous post, I was asking a question.

3). Most Sun employees do not make mistakes 1 and 2.

The third point is what prompts this post. Thanks to some concerned people inside Sun, the person who received the bad information now has a clear line of communication to get it resolved. I hope they do so.

Also, Sun has asked that anyone that feels that they have been given bad advice by a Sun sales representative contact a sales manager to get the straight dope. You can find a list of worldwide contacts on Sun’s worldwide MySQL sales team page.

Sun employees have also assured me that the sales rep mentioned in my previous post was incorrect, and was giving advice contrary to Sun’s policy. It seems clear that some re-training or better training is in order. I hope Sun takes this opportunity to better educate their sales staff.

Sun, Is GPL FUD A Standard Business Practice?

As The Clash reminded us, Know Your Rights!

A good friend whom I trust implicitly recently told me of a disturbing conversation that had taken place between a DBA acquaintance friend of his and a Sun sales engineer. Disturbing enough that it prompts me to remind everyone what your rights and obligations are when you use the community edition of MySQL that is licensed under the GPLv2.

Apparently, this friend of a friend needed some performance tuning done on a MySQL Community Edition installation that spans several servers, and serves up their proprietary, commercial web application. They contacted Sun in order to see about getting some work done on a contractual basis. In other words, they wanted to pay Sun for performance tuning.

They were then contacted by a Sun sales engineer who, upon hearing they have multiple servers running the Community Edition of MySQL serving up a proprietary web app, told them:

1). They were in violation of the license.

2). They would need to release their proprietary app under the GPL, since it uses MySQL client libraries.

3). They could avoid having to GPL their application by paying Sun licensing fees for every server.

Now, I’m not a lawyer. But, in my legally uneducated but GPL-savvy opinion, there is a simple term for this. It’s extortion. “Pay us, or we’ll make sure you have to GPL your proprietary work.” If someone admitted to the bar in any US state wants to correct my opinion that this is extortion, chime in. I may be confusing terms.

However, I am not confused about the rights and obligations granted and ensured by the GPL. Let me cast some light into the darkness created by this Sun salesman.

First, this company uses the Community Edition of MySQL. Therefore, they are bound by the terms of the GPL, and NO other license, be it from Sun or elsewhere. Full stop. Sun can have whatever convoluted license they please for MySQL Enterprise, but if you use the Community Edition you are bound solely by the GPL.

The GPL ensures that anyone that releases code that touches GPL code must also use the GPL. The key word here is “releases.” Let’s give two examples.

First, Linksys recently settled a lawsuit brought against them by the Software Freedom Law Center. They use Linux kernel code in some of their router firmware, and had not released the source code to that firmware. Why did they have to? Because the firmware shipped to end users. They released code to the public in binary form that touches the GPL Linux kernel. Thus, they need to GPL their firmware.

Second, it’s no secret that Google uses MySQL extensively. So why aren’t GMail, Google Reader, Google News, or other such services released to the public under the GPL? It’s simple. They aren’t released at all! These are proprietary applications that are not released in either binary or source form. If your code touches GPL code, but you don’t release your code in binary or source form, you do not have to use the GPL for your code, nor do you run afoul of the GPL itself.

So, let’s go over the three points the Sun sales engineer made above.

1). They were not in violation of the license of the Community Edition MySQL server software they were using. To imply differently is at best misinformed, and at worst FUD.

2). They don’t have to release their code at all! Now, if they do decide to release their code, because it touches the MySQL client libraries they will have to use the GPL. But if they keep it a closed-source, proprietary, commerical app, with no public releases, they are well within their rights. Their application, developed in-house and run as a service on top of the MySQL Community Edition, is not released to the public in any form. Therefore, they do not have to use the GPL with their application or service, nor do they have to acquire commercial licenses from Sun. End of story.

3). They wanted to pay Sun for performance tuning. Sun decided to go for a bigger paycheck. In the end, Sun probably will not get a dime. Smart. Really, really smart. Glad I don’t own stock.

If Sun contacts you or someone you know and begins talking about license violations, read this post again carefully. Know your rights. And if you’re not sure, contact the Software Freedom Law Center. Or contact us at Monty Program; our staff was responsible for putting MySQL under the GPL. We know the terrain better than Sun.

If you have had a similar experience, I’d sure like to know about it. I’m curious as to how widespread this “tactic” is, and how people are responding to it and dealing with it. Let me know via comments.

And while we’re at it, if you need MySQL or MariaDB performance tuning, why contact a company that instead of seeing an opportunity to make a client happy while earning money instead sees an opportunity to extort money from a heretofore satisfied user? Contact the Open Database Alliance. We have people that can get the job done, and that don’t care about your licensing.

OpenBSD SMTPD

The OpenBSD hackers are writing their own SMTPD. Anyone that says “what an absolute waste of time” has:

1). Never run Sendmail.
2). Doesn’t care if their SMTPD is not GPL-compatible (Postfix ain’t).
3). Too much free time and not enough experience.

I’m psyched for this. I’d love to see an SMTPD that

1). Has configuration files that need not be treated as binaries.
2). Is resource-friendly.
3). Is sysadmin-friendly.
4). Is secure.

In my experience, the OpenBSD crowd is among the most likely subset of Free Software and open source hackers able and willing to deliver on this wishlist.

Make me happy, guys!

The Light At The End Of The Tunnel May Not Be An Oncoming Train

Today we at Monty Program Ab and our dear friends at Percona announced the Open Database Alliance. Click the name for the full press release.

If you are a MySQL hacker, user, documentation writer, or just an interested bystander, we’re ensuring the database software you have come to know and love, and with which you (quite rightly) feel a sense of shared ownership, stays Free and open. Not only are we going to keep the same development and participation model with which you are familiar, but we are going to open the project even wider. Got patches? We want them! Got questions, comments, or other issues? We want to hear them, and we will respond in meaningful ways! Got production deployments of MySQL? We will keep you up and running without labyrinthine licensing and high expense! We’re lowering the bar for participation and raising the bar for what a Free and open database can do.

And the point of the Alliance is that we’re not doing it alone. The MySQL ecosystem has a large number of stakeholders, and we want them all to consider the value in being part of a community of like-minded individuals and organizations. The Alliance is here to ensure that your transition to MariaDB is painless, and offers you a more robust and healthy world of code.

We’re just getting started with the Open Database Alliance. If you have built a business or beloved hobby on MySQL, I encourage you to contact us and join the Alliance. We’re here for you if you’re here with us.

Think positively. With the Oracle acquisition of Sun, doors may close. But we have just removed a wall from the side of the house. In Free and open source software, doors are for chumps. 🙂

Call 1-800-THIS-IS-THE-21ST-CENTURY

The days of asking people to dial a phone number by associating letters in your nifty, branded phone number (I’m looking at you, Key Bank and Compass Bank) to digits on my telephone keypad are over. Why? I have a Blackberry Curve. Here’s my keypad:

curve-keyboard

My wife has the Blackberry Pearl. Here’s her keyboard:

pearl-keyboard

Uhhhhhh ….

Asking me to dial 1-800-KEY2YOU or 1-800-COMPASS without giving me the actual digits basically shows your entire marketing team thinks this is the 1950s, and all telephone handsets look and behave exactly alike.

In our home, we only have cell phones. The only POTS phones we own are packed in a box in the garage. I could look up an old-fashioned telephone keypad with a Google image search, but … seriously? And what if I’m in my car? Pull over and find a pay telephone so that I can figure out what number to dial? Ridiculous.

Wake up, suits!

Super Troopers?

Seen the movie? Very funny fiction about the Vermont State Police.

So a week or so ago I was pulled over on I-89 in northern Vermont for speeding. The officer exhibited behavior and judgment so poor and so unprofessional I will not sully the reputation of his peers by discussing it here. I have raised my concerns with his supervisor.

However, that does not negate the fact I was, indeed, traveling at a rate of speed higher than the posted speed limit. So I’ll pay the ticket. And when I went to do so, I noticed that “You can now pay your tickets online at the VT Judiciary Branch website.” Click the URL for some head-scratchin’ fun!

I’m aghast. WTF are they smoking in Vermont?

Vermont Judiciary webpage

The Delphic Oracle Embraces the Sun-God Apollo

Big events today, and the MySQL User Conference is all abuzz with news, speculation, and even some concern. Oracle has acquired Sun, which again brings the light of Apollo to Mount Parnassus. 🙂

Of course, this raises all kinds of questions about the whys and hows. I’m happy to publicly speculate, as I am not in an insider’s position with either Sun or Oracle. Of course, I do work for Monty Program Ab, which makes me somewhat less than a wholly dispassionate bystander. And please, before you read my humble musings on the subject, please read my boss’ thoughts.

Without further ado, here some possible reasons I can see for this move by Oracle:

– Oracle now has a hardware business, something I think they have craved for a very long time.

– Oracle now has their very own trusted server operating system. Not to disparage what must have been some very hard work, but Oracle’s Linux platform has been and will be an also-ran. I predict that Solaris as a commercial product will soon disappear, and that Oracle will concentrate their efforts on OpenSolaris. I think Oracle wants an operating system, but I (perhaps naively) believe they see the folly in closed, commercial OSes in 2009.

– And of course, MySQL. I think it almost certain that we will begin to see MySQL used as a sort of “gateway drug” to Oracle’s flagship DB products.

Of course, this begs the question “Why would Oracle continue to pay attention to and improve a product that is solely used as a carrot to their flagship product?”

I’m not quite sure I know the answer to this. However, I do know that it is nigh impossible to “own” a Free or open source project in the same sense that Oracle “owns” their database. Patrick Galbraith spoke to this point quite eloquently in a recent blog post. I think Monty paraphrased it perfectly when he said “that ‘ownership’ of Free and open source projects has more to do with who provides the best stewardship of the code, rather than who owns a trademark.” Amen, brother.

That having been said, I hope Oracle realizes that they have taken their first big step into the Free and open source software world with their acquisition of Sun. Their track record in this field is non-existent, as they have traditionally been quite opposed to the ideals and values that make Free and open source software viable. They now stand at a crossroads. They can continue to play the closed, proprietary, commercial game they have been playing (quite successfully, mind you) for many years. However, if they do this, they almost guarantee the success of competitors (including Monty Program Ab). I cannot say I’m opposed to this, but it decreases users’ choice, and this is never good.

Alternatively, Oracle could choose to engage and interact with the open source community in ways that Sun never did (e.g. viewing each developer with meaningful patches as a full partner in their success). Should they choose this route, I have every confidence that they will succeed where Sun did not.

When Sun was the principle owner of MySQL, Monty liked to think of Monty Program and the Maria Project as the MySQL Fedora to Sun’s RHEL. Now that Oracle is in charge, I see three possibilities:

1). The Maria Project continues to be the MySQL Fedora to Oracle’s RHEL.

2). Oracle completely fails to understand the open model and Maria becomes the Linux to Oracle’s Windows.

3). Oracle really and truly understands the landscape in which they now live, and the Maria Project becomes the Ubuntu to Oracle’s Debian.

As I said, there is palpable concern amongst the conference attendees. I think there are some easy tasks Oracle could undertake to alleviate these concerns to some degree. I’ll post my ideas in the next few days, once I’ve a a bit more time to process today’s events.

Fasten your seatbelts and return your tray table and seat back to their upright and locked positions.