MySQL CEO Mårten Mickos Answers Your Questions

You asked. Mårten answered. He even added (and answered) a question he wished had been asked, but wasn't. If you have a comment or follow-up question, please post it. Mårten will spend as much time responding to your comments as his schedule permits. 1) Re:Biggest Problem?
by wild_berry

In light of comments here about PostgreSQL being superior (and F/LOSS) technology but MySQL being used because of "MCSE weenie mentality", are you, Marten Mickos, concerned about the complacency of 'good enough' technological solutions or that Free/Libre Open Source Software may never remove the entrenched market leaders?

Mårten: Not the least. There will always be fanatical supporters of various products and projects, but I would look at what real users are saying. As an example, Nortel, Alcatel and Nokia are building new mobile phone network elements that run on MySQL Cluster. That's leading edge open source technology in some of the most advanced high-availability use. Or take YouTube or some of the other new web successes - they are scaling incredibly fast and again it is open source that powers them (MySQL, to be specific).

It is a common misperception that advanced technology cannot be easy to use. But it can! Driving a Tesla is fairly straightforward, but building one is not. Flying the new Eclipse jet is comparatively easy, but designing and constructing the aircraft is extremely difficult. And although it was pretty difficult to fly Spaceship One out of the atmosphere twice in two weeks, it was nothing compared to the difficulty of designing such a modern spaceship.

2) Perception of low quality for 'free'
by OakDragon

How do you fight the perception that MySQL is not suitable for 'the real world' because it is free?

Mårten: That's a challenge we share with all open source products, and I believe that such misperceptions will be removed over time by the overwhelming momentum of open source at large. So it is not so much a product-specific issue. But we need everyone's help to set records straight.

3) R&D Directions?
by eldavojohn

In a market where people are just looking for stability, simplicity & scalability, where do you turn for innovation in your products? Is there a lot of research and development towards new features and completely new products in MySQL's community or do you aim primarily to do one thing well? How do you influence the direction of this research in such a large open source project? Do you attempt to add direction at all?

Mårten: I am very proud of the innovative power inside our company, but even more important is the notion that "innovation happens elsewhere" (that's a great book that I recommend, as is "Democratizing Innovation"). We just announced a new monitoring and advisory service which we have innovated inside the company and with the help of customers.

But more massive is the innovation that happens in the MySQL ecosystem - in the user space. Look at the XML wrappers, RSS converters, full-text search modules or any of the toolkits that have been developed for MySQL. Or take PBXT - a new transactional storage engine, or the custom-built storage engines that Google, Friendster and others have developed. It is in the interaction with these advanced users that we learn what's useful and what isn't.

It takes a degree of humility and openness (and eradicating any Not Invented Here feelings) to be able to make good use of innovations from the ecosystem. But in return it is a much more productive way. I see it as a Darwinian system where the best ideas survive. Nobody can know for sure exactly where the development is heading, but if millions of MySQL users try out various new ideas, all the right things get developed.

We still need to improve our ability to receive innovations from the community, but I am proud of how far we have come already. Look at the MySQL Forge if you are interested: fforge.mysql.com

4) Re:R&D Directions?
by bzipitidoo

A few questions, only one about the biz. For those of us who like designing and researching and wish to spend our time on that, and don't like the thought of spending time on sales and promotion, or begging for money from VC vultures, or figuring out tax forms, stock options, and such, or sifting through thousands of resumes trying to find a few good people to hire, or knowing when a deal is a bad one to be avoided at all costs, or all the other aspects of business, what are wannabe independent software developers to do? If I try to start a business, it'd be so I can make some money doing what I like, not sink time into the business of business. How'd MySQL handle that when it started?

The parent's R&D question reminded me of another question. A goal of future versions of ReiserFS is to make database programs unnecessary and obsolete. Reiser asserts the only reason you even need a database program is because current file systems cannot do things like efficiently store, access, and query thousands of small records. File systems should be able to do everything a database can do. Apart from the bit about being in jail, is Reiser crazy, or is he on to something?

Anyway, do you recommend or favor some file systems over others for best performance from MySQL?

Mårten: I think you had a total of three questions.

First, when MySQL got going, it initially was run day-to-day by the three founders Monty, David and Allan. In the year 2000 they realised it was growing beyond what they could handle, so they reached out to get a CEO, a board of directors, and VC capital. This is a great model for those who want the company to grow and who are ready to share decision-making with others.

Your other question is about what a software developer of today could do. I believe it is easier to start a business today than 10 years ago, and I believe there are some great avenues to follow. For instance, you can develop open source software that wins the hearts and minds of users, and then you can license or sell the software to a bigger vendor if you want to avoid the hassles you mention.

Mark Matthews is a great example. He had developed the best JDBC driver for MySQL, and then we acquired his software and hired him full-time. JBoss has done similar "acquisitions". Another alternative is to build a website that provides the intended functionality as a service. The benefit there is that you don't have to build a distribution channel, because the web is the channel. And then there can be some great outcomes if you don't want to run your own company. Just look at Flickr or YouTube, for instance, who got acquired by web giants for good sums of money.

Your third question is about filesystems versus databases. If I were a filesystem developer, I might claim that databases will become obsolete. But as I am a database developer, I will claim that filesystems will become obsolete! Humour aside, I don't have a specific recommendation, but I do believe that we will always have heterogeneity in the software world. There will always be several alternative technologies for every single problem.

5) Conflict of Interest
by 99BottlesOfBeerInMyF

One of the most common complaints I've heard about the business model of profiting on support for a product, is that it provides motivation to keep the product from becoming very user friendly. After all, if the product is too easy to use, who will pay for support? In my own experience, I've seen a lot of companies that consider support to be insurance, and don't use it for help with installation, configuration, or to overcome usability issues so much as a way to cover their asses in case something goes very wrong. Do a lot of your customers use support to overcome usability problems and if so, does this de-motivate you to solve other usability issues?

Mårten: This is a great question. First, I think closed source vendors are proving the hypothesis incorrect, because they are the ones who have un-friendly products although they have a licence revenue stream. And open source products, which typically lack a licence fee, are the ones with the best user friendliness. Why is that? I think the reason is that popularity is worth more than the marginally improved fees you could get for a user-unfriendly product. Sure, we lose some purchase orders because customers simply don't need our support. But at the same time we gain millions of new users, and they in turn help us develop the product and drive our marketing.

I think one of the reasons MySQL is so popular is that we continue to take user-friendliness seriously. We may joke about leaving some bugs in the product in order to have more support fees, but in reality we take the bugs very seriously. And when we at times have more bugs than we would like to have, the whole organisation is in pain until we get the upper hand and kill more bugs than we produce. Version 5.0 was such an example - when it first came out as GA it had more bugs than earlier versions. It didn't feel good to see the list of incoming bugs, so we rearranged tasks and expanded our QA test coverage in order to fix the situation. I hope you who have used the latest versions of MySQL 5.0.26 have seen the significant improvements.

6) Defects per KLOC
by eldavojohn

Your website touts you as having the lowest defects per KLOC by up to 12 times the industry standard, what do you attribute as the leading factor to your success in this respect? Since cold cash is the traditional method, how do you incentivise code quality in an open source product?

Mårten: I think open source has a wonderful benefit over closed source that often is forgotten: When you know that your work may be scrutinised in public by others, you simply do a better job. It is some magical combination of pride and (fear of) shame.

So I am claiming that the same developers will do a better job of producing open source than closed source software. The analogy is your backyard: why is the backyard never as tidy as the frontyard of a house? You spend more time in the backyard, don't you? And it is you taking care of both the front and the back. The only difference is that your neighbours see your frontyard but not your backyard. It's as simple as that.

Additionally, the reality of the scrutiny, i.e. the peer reviews, ensure that a lot of bugs and defects are detected and removed early. So I am not the least surprised that open source is 12 times cleaner than closed source. The really scary thing is all the stuff that is buried deep inside closed source software, out of the sight of everyone and without any proper stewardship.

7) MySQL business model for niche products?
by MarkWatson

Hello Marten,

First, congratulations on MySQL's market capitalization! My question is:

I have been working part time for about 6 years on software for text/data mining and general semantic information extraction. Almost all of my development is in Common Lisp, but I have ported little bits to Java and released that under the GPL in the past. I view this as a small, niche market, not like MySQL. What do you think that chances are for making money on GPLing a niche product?

MySQL is very widely used so if you capture commercial use icensing costs for a small percent of users, you do very well. For my software, with luck perhaps a few hundred companies a year might start adopting my product. Does it seem like wishful thinking for me to use a GPL based business model like MySQL's?

I want my customers to have my source code for a lot of reasons, but I would also like to capture revenue. I might just end up going to market as a proprietary product that incidently includes source code, with licensing that prohibits redistribution to non-customers.

Thanks for your help,

Mark

Mårten: Good question. I used to state in presentations that open source probably is best suited for products with huge user bases. But I was always proved wrong. Somebody would come up to me and tell about a successful open source niche product. So I have learned that you can be successful with open source in any environment.

I think the key is to provide real value to your users or customers - initially free of charge, but for continuous support against a fee. In that way you can build credibility in your market and gain important insights in what customers need, and you can also build a business. But don't think it is easy to find a business model that fits with free or open source software. It is difficult. But it is possible. You have to take some risks before hitting on the right answer.

8) Commercial vs free - where to draw the line
by Internet Ninja
I work for a good sized business and looked at using the Cluster Jumpstart but when I told my boss the cost and that we'd have to pay for flights he laughed at me, even though we're starting to really use mySQL pretty seriously now for some stuff.

With costs for things like this and gold/platinum support also relatively high on a per server basis it seems there's a wide gap between community based support which costs nothing and enterprise support which appears somewhat pricey.

How do you draw the line for paid vs free support particularly since a lot of SME's are using mySQL and may be unable to afford it? Was it a conscious decision to pitch it high to display value in the product?

Mårten: This is essentially a question of price elasticity. If we lower the unit prices, we may sell more, but will we sell that much more? If we increase the prices, we may have fewer customers but they may be more profitable.

We have an ambition to be affordable and available for all, and therefore our prices are typically a tenth of what you would pay for closed source databases. But MySQL Cluster is a state-of-the art solution that in its own market segment beats all other databases hands-down. The price may sound high to someone from a smaller company, but it is carefully weighed against market conditions. Many customers find that it is not only the best solution technically, it is also the least expensive one. If the price is prohibitively high for someone, then they can always use the community edition and do the work themselves.

9) Achievements & Fallout
by eldavojohn

In your five years as MySQL CEO, what has been your proudest moment? Do you find it difficult to lead a company based on a product that belongs to a community? Do you ever experience any fallout/backfire from running your company on such a business model?

Mårten: Thanks for the question! I felt very proud when we visited Google for the first time and people would come out of their rooms just to see the "MySQL folks". I had scheduled a CEO-to-CEO meeting with Eric Schmidt, but in addition to him we had a dozen developers who all wanted to know about MySQL internals. Fortunately, I had David Axmark (co-founder) with me and he did the technical presentation! Another proud moment was our users conference this year with a thousand or so in the audience, all gathered just to listen to the State of MySQL. Those situations make us realise that we are changing the world - we are making a difference.

The great thing with our business is that we get to pursue goals that we are passionate about. We get to work with the smartest people in the world on the newest and coolest database technologies. We get to stand up to defend the freedom of software and to fight software patents. We contribute to the wealth and happiness of what is called the bottom of the pyramid - people who may not have the financial means to buy software but who have dreams and goals they want to pursue. Of course it all needs to make sense from a financial perspective for everyone involved, so it all needs to boil down to a good business model. But, as I noted, that is exactly the great thing with our model.

But we also need to know that it is very difficult to walk the fine line between freedom of software and freedom to pursue profits. The two are not at odds with each other, but there are overlapping areas where you need to have all details right. We have made our mistakes over the years, and I hope we have fixed them. If we haven't, let me know and we'll get onto it. I am not expecting everyone to agree with our business model, and every now and then we get flamed by someone, but I do hope that people generally respect our business decisions. And so far that has clearly been the case.

A friend of mine just wrote to me: "If you can find a new way for folks to make money from FOSS without compromising fundamentals, good for you. Others can follow your lead. Just don't compromise the fundamentals." That captures the essence of the question very well.

10) Appliance possibility?
by Roblimo (added editorially
because of this discussion)

Now and then we hear rumors that Oracle is going to either come out with its own Linux distribution, sell a single "stack" with Oracle products running on Linux or else sell a complete hardware appliance that runs Oracle on Linux.

Back in 2003 there was a significant PR splash touting database server appliances made in partnership by Pogo Linux and MySQL, but we haven't heard much about that idea since. We know you now have many channel partners, including hardware vendors, but this is not quite the same.

Do you have any plans to come out with either a Linux/MySQL "single stack" software product or a preloaded Linux/MySQL+hardware appliance? And if not, would you change your mind if Oracle started selling either of these products?

Mårten: First, if Oracle launches their own Linux distro, then we take that as a great victory for open source. (And we ask - is the database next?) The most successful distros include JBoss and MySQL and many other products, so we would naturally make sure that Oracle gets all the help they need in putting our product on their operating system.

As for appliances and stacks, I am not sure that we would choose to compile our own stack. There are so many tempting business models and products, but to really grow fast there are just three main rules: Focus, Focus, and Focus. In our ecosystem we have many partners who could provide a stack, and we are very supportive of them. HP has the Linux reference architecture, Pogo their appliance, and Sourcelabs and Spikesource came out with software stacks. The power of open source and the power of LAMP is in the collaboration between the companies, not so much in what one single company can accomplish.

11) What about community?
(inserted by martenmickos himself)

MySQL is enormously popular and has a vibrant ecosystem, but it seems that the core product is primarily developed in-house. Are you at all interested in contributions, and what are you doing to stimulate community engagement?

Mårten (answering his own question): Initially, although open source, MySQL was developed by just one person - our founder Monty. Over the years we have expanded the group of developers both in-house and in the community. But we have more work to do. We need to make the product more modular so that it is easier to contribute distinct modules (such as storage engines or full-text search modules). We need to document the source code better (we are doing it with Doxygen). We need to make the contribution process as easy as possible.

We have very strict requirements on quality and ownership of intellectual property before we put something in the commercial product, but in the MySQL Community Server we are becoming much more open to contributions. We also need to set up websites and systems for broad collaboration in our community.

Planet MySQL has become a great aggregator of MySQL blogs, and the MySQL Meetups are popular local events. The MySQL Forge is a place for contributions, and soon we will launch the Winter of Code program with coding contests. For the most passionate MySQL users and developers there is the MySQL Camp in November (mysqlcamp.org). And of course all of us will get together in Silicon Valley in April for the annual users conference.

But no matter what we organise on behalf of MySQL AB (by the way, AB means "Aktiebolaget" i.e. "Incorporated"), the true value of the community will always be generated by the community itself. We are still on a learning curve in terms of all the possibilities that this offers. So please share your ideas with us. What should we do more of and what less? How could we bring together MySQL users from all over the world to accomplish bigger things? How could we together change the world for the better?

It is amazing what open source has accomplished in the last 15 years, and I am even more excited about what it can do in the next 15. But my excitement is anchored in our community. Alone I can accomplish nothing, but with the help of millions of users worldwide, we can accomplish just about anything.

I hope I answered all questions clearly and conclusively. Let me know if I missed anything.

Marten

------

Comments

To serve Man.

[Anonymous Coward]

"If you have a comment or follow-up question, please post it. Mårten will spend as much time responding to your comments as his schedule permits. "

Hopefully we'll treat this interviewee better than the last one.

Re:To serve Man.

[martenmickos]

Thanks. I am ready for both easy and tough questions and look forward to a healthy debate.
 
Marten

Re:To serve Man.

[Espectr0]

To what do you attribute your success over Postgresql, which supported true ACID and more important features like subqueries, triggers and such a lot earlier than MySQL? Do you think it's because you provided a Windows version and they didn't?

Re:To serve Man.

[martenmickos]

In my experience, success is always a result of a number of things that are done right, plus an amount of luck. In other words, many stars must be aligned.
 
I believe that for MySQL it was vital that the founders were so customer-focused from day one. They responded to every single email. They listened carefully and they took care not to be arrogant. They wrote a good reference manual. They made sure the product could be downloaded and installed in no more than 15 minutes (and you may have noticed that the best programmers are usually also impatient, and you lose them unless you provide value to them quickly).
 
Marten

Re:To serve Man.

[Anonymous Coward]

Hopefully we'll treat this interviewee better than the last one.

The last guy got roasted because he tossed off some glib & patronising answers and proceded to act like a dick in his subsequent comments.
By contrast, Marten has been respectful and insightful in all his answers. I would be very surprised to see him get widely flamed.

Re:To serve Man.

[nuzak]

MySQL was much easier to install and ran faster than Postgres at the time (this also largely explains the success of PHP -- mod_perl is *still* a pain to work with). In fact, when it comes to things like replication, MySQL is still easier to install -- Postgres-R and Slony-II always seem to be hiding off in a corner, away from any sense of being "official". Since MySQL's primary niche was and still is to back web apps with a single front end UI, API, and usually a small development team (typically one person), having to put basic database logic like transactions into the application layer wasn't too punitive to preclude considering mysql.

Mind you, I still remember with disgust how MySQL's docs would trash the whole concept of transactions, and I still think it's abominable how MySQL's own system tables, very critical things, are still stored in MyISAM, and how the whole MyISAM storage engine needs to die, etc ... but it really doesn't matter to many shops. It's quite clear that the trend is to move business logic like triggers and SP's out of the database (transactions do of course need to stay in) and move to a generic storage engine. MySQL was born into and is growing up in that niche, while other databases are adapting themselves to it.

In short, its success can be summed up in three words: First Mover Advantage

We have many terabytes of data and all kinds of complex business logic where I work, and it's all in Oracle, but there's hardly a SP to be seen. Some functions and views are there for convenience, that's about it. And I'm finding that this is not an unusual setup. For my own apps, MySQL still isn't being considered, but that's largely for the misfeatures (concerning data integrity) it has, not the features it lacks.

Re:To serve Man.

[Amazing Quantum Man]

How much, if any, has your strategic partnership with SCO hurt MySQL, either financially or image-wise?

Re:To serve Man.

[Anonymous Coward]

Hopefully we'll treat this interviewee better than the last one.

We're sorry, Beckerman.

Re:To serve Man.

[martenmickos]

Thanks for the question. A detail: the partnership is a regular one, not a strategic one (where "strategic" would mean that it is a top priority for the company and we have a whole team working on it, like we do with HP or SAP for instance).
 
We certainly have been criticised for striking a deal with SCO, but I think we also have explained our reasoning and motivations well (especially on Groklaw a year ago). Financially we have gained from it because there are payments from SCO to us. To reiterate our main motivations: there are end-users on SCOs platform that we think deserve the best database they can get, so we think it is right to serve them. We do not agree with SCO on what they are doing in the lawsuits, but then we also think we should not be the judge of that. The customers of SCO can vote with their feet, and the courts of law are there to pass the verdicts. Please read the full Groklaw discussions for more on this.
 
What I hope the world sees is that MySQL AB makes up its own mind (even when it may not be the most popular decision), stands for its decisions, and is ready to openly debate the pros and cons of any decision. In the long run, that's in my mind the only path to success. We may make wrong decisions at times, but at least we are making decisions and at least we are moving forward.
 
Marten

Re:To serve Man.

[Chapter80]

To what do you attribute your success over Postgresql?

The fact that one can say MySQL without spitting up their free (as in beer) beer.

Re:To serve Man.

[Kalak]

I'm replying to this solely to remove the accidental "-1 Troll" mod I gave it. I did not in any way view this as a troll, and how do you troll yourself anyway?

I need to double check before I mod. I'm glad I caught it after submitting it so I can revert it.

Re:To serve Man.

[evought]

Sorry I am late to the game with a question, but:

I met you several times while you worked with SolidTech. I was a happy user of Solid myself and through several clients. In particular, it made a great development database since it was lightweight and highly standardized (API-wise).

I found it easy to get small clients to invest in Solid for deployment which made it easier for me. I had trouble getting larger clients to use it for either development or deployment. They gnerally said the licenses were 'expensive' and then went and bought Oracle. Over time, as Solid Server disappeared, I moved to PostgreSQL.

Anyway, my question is, what made you go from Solid to MySQL and what do you think was Solid Server's failure? Was it just the competition from the FLOSS alternatives, or was there something more?

Re:To serve Man.

[martenmickos]

In some ways, the question is by now nearly ten years old, but here is my story: At Solid I was in charge of a campaign to make it popular on Linux which was a novel thing at the time. For a very short time Solid indeed was the most popular Linux database (in 1996 or so). So when Solid's board decided to focus on big opportunities in telco and telematics rather than on popularity on Linux, I decided to leave.
 
That's one of the reasons why I a few years later so gladly accepted the offer to become CEO of MySQL - I could take up the idea where I had left off att Solid, and with a momentum a hundred or thousand times stronger.
 
Interestingly, when we fast forward to today, Solid is back in this game and has just released their open source version which integrates as a storage engine for MySQL!
 
Marten

Re:To serve Man.

[merc]

What I hope the world sees is that MySQL AB makes up its own mind

Of course MySQL AB has the freedom to make up its own mind. As a consumer I have several criteria that I can use when I decide to spend my corporate dollars. Some of us are offended by your partnership with SCO/Canopy and have chosen to boycott your product despite any technical merit it may or may not have. Freedom works both ways, and in my view MySQL AB caused themselves irreparable damage when they decided to partner themselves with villains.

Re:To serve Man.

[martenmickos]

Yes, you of course have the right to stop using MySQL just as you have a right to start using it. I am sorry if we have lost a customer in you. I respect you for taking the time to tell us about it.
 
You may not be interested in any further details, but if you are, feel free to send me an email and I would be happy to set up a time to discuss this over the phone.
 
Marten

My Perspective on MySQL

[laptop006]

I know three local(ish) MySQL employees, and they're all great people who've been pretty much the best at what they do.

So shouts to Arjen Lentz, Stewert Smith and Colin Charles.

Comment about this interview

[Anonymous Coward]

The MySQL CEO's responses are very fast, but he still lacks ACID compliance.

Re:Comment about this interview

[Jay Pipes]

Hmm, you must be referring to Marten's interview from before *2001*, then? ;)

Re:Comment about this interview

[aminorex]

Indeed, the hippies will never be satisfied until every CEO in the world sees the light through ACID.

Thanks

[eldavojohn]

If you have a comment or follow-up question, please post it. Mårten will spend as much time responding to your comments as his schedule permits.

I have one comment: Good answers to the questions. And thank you for your valuable time! It's nice to know that MySQL has an intelligent respectable CEO. All too often we concentrate on the Patricia Dunns of the world, it's people like you that make me second guess my claim that I would never want to be CEO of a company.

On that note, one last question I had though of would be, "Do you view yourself as a typical CEO or an abnormality in the pay-band of sharks that upper corporate America so often seems to be?"

Thanks again!

Re:Thanks

[martenmickos]

Thanks for the question. I don't know what kind of CEO I am. I am just following the instincts I have developed over the years - from being a boyscout when I was young to engaging in all kinds of business and non-business activities in my late teens and twenties, and later. Most of the time I have chosen to do what is fun rather than what could have brought fortunes, so perhaps I am different in that sense. But as CEO my task absolutely *is* to create fortunes. Perhaps you should ask people who have worked with me or work with me. They may have a better perspective.
 
Marten

Re:Thanks

[Anonymous Coward]

Blast from the past, Mårten, inquiring minds want to know: do you still enjoy an occasional shot of Carillo??

Re:Thanks

[martenmickos]

Carillo, Carillo, Carillo - siinä maistuupi mansikkahillo!
 
Marten

Thank you for the responses...

[x-vere]

These are good responses. I, as a reader, appreciate the time you put into this. I have been a fan of MySQL for a couple years now and really appreciate the product. In my circle, MySQL 5 has come out as a serious competitor to SQL Server because the few thing we needed in SQL Server (stored procedures) can now be done in MySQL. This is fueling the considerations of embracing more open source products in our organization.

Re:Thank you for the responses...

[Shados]

If the only need you needed from SQL Server was stored procedures, there was a ton of alternatives a long time ago. And stored procedures are (to some extent) being phased out by other things. That being said, when you set up SQL Server, Oracle, etc, its not for a feature from 10 years+ ago. Its for things like Business Intelligence service integration, data mining, etc.

Re:Thank you for the responses...

[x-vere]

It was only an example. However, another reason to stick with SQL over other things is skill sets available, rigidity and comfort zone of decision makers etc. I should have realized the need for greater detail as the inevitable did in fact happen. Someone felt the need to flex their muscles.
In my experience, smart choices are often passed up for easier choices.

Competition from above and below.

[Anonymous Coward]

MySQL is facing extreme competition from above and below. PostgreSQL is currently expanding both upwards towards Oracle and DB2, while concurrently expanding downwards towards MySQL. On the other end, SQLite is rapidly putting upwards pressure on MySQL. We find that both of these alternatives have a much friendlier licensing scheme than MySQL, and both have features and quality that MySQL currently lacks.

How will MySQL respond to these threats? With SQLite being a suitable replacement for low-end tasks, and PostgreSQL proving excellent for anything beyond what SQLite can't do, is there really any need for MySQL these days?

It's no doubt that people will continue to use MySQL for years, just due to their current investments in existing MySQL installations, as well as the time they spent learning MySQL. But beyond that, does MySQL have a future?

Re:Competition from above and below.

[KZigurs]

Yes, it has. It has the best windows installer (don't laught. This is THE reason.)

Also - been to computer book store recently?

Re:Competition from above and below.

[martenmickos]

In free software and open source communities it is often believed that MySQL's competition is Firebird, Postgres, and SQLite. But I see it differently. Those products are of great quality and they will definitely continue to have strong followings (that may very well grow). Taken as a group, we compete with each other for the hearts and minds of open source developers, and thereby we also (each one of us) become better over time.

But elsewere it is MySQL that creates the competition. We are successfully scoring victories over the "big three": Oracle, Microsoft, and IBM. That's where the main battle is being fought. We draw strength from our enormous user base in the FOSS community and use that to compete in the commercial world. That market is estimated to be 15 billion USD. Many of you may not care about business and dollars or euros, but as CEO I do, and I know that by winning in the commercial market, we get money for which we can develop more FOSS software that you then can use and modify and re-distribute and so on.
 
Make sense?
 
You also have a comment on our licensing scheme. My view is that it is unfavourable only for the ones who are trying to close-source their apps without giving anything back. If you are all FOSS (Free and Open Source Software), you can use our software completely freely. If you wan't to close something, we will be happy to sell you a closed licence. In summary, we have a business model that works, which is what allows us to grow the organisation, support FSF, fight software patents. etc. But I continue to be interested in your suggestions on how we can improve our model. Have a look at the MySQL Forge and our new program for encouraging congtributions, for instance. Does that make sense to you and is it a change to the better? If not, what would be? Thanks!
 
Marten

Re:Competition from above and below.

[Ant P.]

The OP is spot on. Besides being easier to install, MySQL has things like the PHP MySQLi interface. SQLite actually has a better PHP interface IMO, but sometimes it's literally impossible to get things done with it.

Re:Competition from above and below.

[julesh]

PostgreSQL is currently expanding both upwards towards Oracle and DB2, while concurrently expanding downwards towards MySQL. On the other end, SQLite is rapidly putting upwards pressure on MySQL. We find that both of these alternatives have a much friendlier licensing scheme than MySQL, and both have features and quality that MySQL currently lacks.

Err... while I buy the fact that pgsql (as a BSD licensed product) has friendlier licensing than MySQL (GPL with commercial licenses available), SQLite is also GPL, and I don't see any mention of commercial license availability on their site, so I'd say their licensing is *less* friendly.

Re:Competition from above and below.

[nuzak]

Both postgres and mysql use MSI to install on windows, and both install a windows service. Postgres even creates a postgres user account in the installer and installs its service to run as that user, while mysql runs as .\LOCALSYSTEM. Postgres has lots of checkboxes in its installer for the various features you might want, like GIS, RTree, etc, while MySQL has a few nice options for setting defaults based on the expected use pattern.

Speaking as someone who installs both on workstations all the time, I can say the out of the box install experience is pretty much identical.

Re:Competition from above and below.

[nuzak]

> SQLite is also GPL

SQLite is in fact public domain. In jurisdictions that don't recognize the public domain, this tends to devolve to a BSD license without the requirement to distribute the license doc.

Re:Competition from above and below.

[shmlco]

I agree that one of the drawbacks to the licensing schemes is the price, but really, that's unavoidable. In closed-source software, you have thousands, if not millions, of people who pay some small amount for the product and that, in turn, buys down the costs of support, maintenance, and R&D.

In an OSS project like mySQL, almost all of those people are using it for free. And without anyone to buy down the costs mentioned above, paid support costs and licenses must be higher, as the people who are there to provide those services have to be paid.

If you want free software and free/cheap on-demand support and maintenance and free/cheap licenses then I'm afraid you're engaged in wishful thinking. Because in that case even "free" software has costs...

Re:Competition from above and below.

[julesh]

So why is there a copy of the GPL in their CVS?

Re:Competition from above and below.

[julesh]

Ah, ignore me. They're using a bizarre CVS/Web gateway that lists deleted files inline with ones that are still in the current version but with a different icon. And doesn't mention that they're deleted on the log page for them. Strange.

Re:Competition from above and below.

[19061969]

Absolutely correct

Re:Competition from above and below.

[killjoe]

It's taken them a long time but postgres does have a nice windows installer. Now they need to make it easy to set up replication and clustering and get some nice admin tools.

Oh support for GUIDs would be cool too.

Re:Competition from above and below.

[nuzak]

> support for GUIDs would be cool too.

It'd be nice if they packaged it with the installer, but this works: http://www.hclausen.net/psql.php

The nice thing about postgres is how extensible it is. In my own project I'm using an optimized version of the cidr type for IPv4 addresses only (since that's all I support anyway), and it's blisteringly fast. Custom operators and indexes rock (as does Pl/PgPython, though I'm no longer using it in that project)

The admin tools for all the open source databases pretty much suck equally, though there's a brisk aftermarket for mysql admin tools. Haven't ever had to set up replication myself, but I've heard that indeed it's a bear.

Re:Competition from above and below.

[killjoe]

Tghanks for the tip. Postgres is indeed very extendable and very capable. It's just not that easy. Slony is so much harder to set up then mysql replication, slonyII is dead as far as I can see. No clustering in site either.

Nothing much said

[mcrbids]

This is a CEO talking. He is incapable of really talking techie - so it should be no surprise that his answers sound very "fluffy". He is, after all, the CEO, charged with the task of maximizing the public image of his company.

He all but refuses to give advice where he's uniquely qualified to do so. (you have to really read between the lines to see that he does, in fact, answer the question, but in a VERY indirect way) He doesn't really comment much on any tech questions other than to paint pictures of daisies all over the question.

So, he's probably a good CEO - but these answers won't mean much to the tech types that /. is going to attract.

Re:Nothing much said

[The Clockwork Troll]

With the exception of the question about how they achieve such a low defect/KLOC rate, none of the questions really begged particularly technical answers.

Re:Nothing much said

[The Clockwork Troll]

OK, I have to concede he handwaved around the filesystem question pretty blatantly

Re:Nothing much said

[TheFlyingGoat]

I have mod points, but I just had to respond...

Are you freakin kidding me? Despite focusing on the business aspects of each question, these are some of the most straightforward and insightful responses I've seen in a slashdot interview. While he briefly touted a few new services and features, he focused on answering the questions instead of turning it into an infomercial. If a MySQL developer were asked the same questions, you'd definitely get a more technical response, but considering he is a businessman, these responses are excellent. The example people are giving about him sidestepping the filesystem vs database question is one place I thought he gave an exceptionally good answer. Instead of getting into a pissing match with the ReiserFS folks, he basically said that every situation can have multiple solutions, which is exactly the same thing that the programmer's motto TIMTOWTDI states.

Overall, I think he did a very good job in answering every question. If slashdot wants a more technical response, they should ask a lead developer for the interview.

Re:Nothing much said

[Anonymous Coward]

What interview did you read? The questions weren't technical, didn't demand technical replies, and he answered them as straightforward as anyone could hope for. Obviously you have never heard an interview with a run-of-the-mill CEO of a large company, because it doesn't get any better than this.

It's not every day we get an interview with a CEO of a very large and influential company who is willing to jump right into the Slashdot discussion using his own account, but if your attitude towards it prevails, it may well be the last. It's pretty sad that your comment is currently at +5.

Re:Nothing much said

[Anonymous Coward]

Somehow, just like the other AC, I somehow managed to miss the ReiserFS question, and I agree that he answered that one very much in the standard CEO fluff-way. Other than that, though, I think the answers really were quite good.

Re:Nothing much said

[jZnat]

Dude, he has a /. account; you can't get more techy than that. ;p

Re:Nothing much said

[belrick]

Yes you can. You can have a Slashdot account with fewer than 6 digits. :-)

Re:Nothing much said

[WilliamSChips]

I really wish my Slashdot account had four digits right now :P

If I am a newbie..

[lemur3]

What is the best advice/suggestion that a person just starting to use PHP+MySQL could get ?

Re:If I am a newbie..

[lemur3]

...and what recent developments could make it easier to learn/use ?

Re:If I am a newbie..

[KermodeBear]

Two things.

One, make sure you have a strong foundation in OO programming. Yes, I know; PHP isn't strictly OO, but the structure and discipline it takes to write code in an OO language will help you out immensely since you can apply a lot of the same principles.

PHP isn't a great first language. It is extremely loose and it is easy to form bad habits early. But, if you are disciplined, you can rapidly produce good, reusable code.

Second, learn some database theory - especially normalization. A great book that I still refer to is "Database Systems: A Practical Approach to Design, Implementation, and Management" by Thomas Connolly and Carolyn Begg.

So, ultimately, the best advice to a person just starting with PHP/MySQL is to get a good foundation in the fundamentals of programming and database systems.

Once you have that, it doesn't matter what language or DBMS you use.

Re:If I am a newbie..

[Ant P.]

Never put a variable inside a mysql_query().

There's a 99% chance it'll fuck you over with an SQL exploit. And with the other 1%, you still get the query cache rendered useless from storing hundreds of small variations on one or two query strings.

Use prepared statements for anything that doesn't completely fit between a pair of single quotes.

Re:If I am a newbie..

[Anonymous Coward]

The query cache does not work with prepared statements, at all.

Re:If I am a newbie..

[Anonymous Coward]

Replying anonymously because I moderated on this story.

I would qualify your statement somewhat. Never put an untrusted variable into a mysql_query(). Of course you can (and should) put variables into a query, but use mysql_real_escape_string() first.

Your comment about the query cache makes no sense though. If you're making different queries, then the cache won't be used no matter what. The only point I can think of that you're trying to make is that, allowing someone to pass in a string directly means they can change the case, which wouldn't change the result of the query, but will change whether the query cache is used or not. This is a different problem though, which can be solved any number of ways. It's certainly not a critical reason not to pass a variable into a query.

Re:If I am a newbie..

[ShieldW0lf]

ESPECIALLY NORMALIZATION!!!

Of course, once you learn about normalization, you might not want to use MySQL anymore...

I've never been asked about PostgreSQL or SQLite

[brokeninside]

... at a job interview. Been plenty of times where I've been asked about MySQL. I don't think there is any danger of MySQL not having a market.

Just do a search on Monster. 101 job hits come up on Postgres. 3 job hits come up on SQLite. Thousands of hits come up when searching for MySQL.

Beer as in Beer?

[bigattichouse]

With all this discussion of free-as-in-free and free-as-in-beer...

What is your favorite carbonated alcoholic beverage?

Re:Beer as in Beer?

[martenmickos]

I enjoy most of them! Somebody gave me a can of "Martens Lager" from Belgium (a famous beer country) so that's a favourite. I am a great fan of Guinness, especially when I visit London and Dublin. I've come to like Samuel Adams and Anchor Steam in USA. In my native Finland I enjoy Karhu ("bear"). The list is longer but I'll stop here.
 
Marten

Re:Beer as in Beer?

[cerberusss]

Let me guess.... You'd like to drink more, but you don't have the time??

Re:Beer as in Beer?

[raddan]

Being from Boston, I am obligated to like Sam Adams. But the locals really like Harpoon. If you care for IPAs, you need to try Harpoon IPA the next time you're around here.

Thanks for the great interview. Just finished setting up another MySQL DB just this morning. I suggest you pick up a Duchesse de Bourgogne (Flemish beer) as a reward for all your hard work :^)

Re:Beer as in Beer?

[Anonymous Coward]

Anchor Steam =). This is my kinda CEO!!!

Re:Beer as in Beer?

[19061969]

Try Leffe - one of the best beers in the world (and I've tried a few!)

I thought his answers were very direct

[brokeninside]

He's a CEO and he spoke to the business side of each question. For example, look at his question about whether MySQL was going to get into the database appliance market. He said point blank that MySQL AB is going to focus on their engine and let others do the integrating of software/hardware. That is a direct, non-nonsense answer.

Further, many of the questions were very poorly formed, especially in light of the fact that they were being directed to someone running a business rather than writing code. For example, the question of whether databases will be deprecated in favor of file systems. From a business perspective that is a meaningless question as the functionality remains the same. It makes no difference to a company like MySQL whether they're selling their engine as a MySQL DB or MySQL FS. He even hinted at the lack of differentiation in his answer that of course the fs folks say that the db will go away but we see it as the fs going away. If the fs and db merge then it follows that db vendors become fs vendors and vice versa. Its not really a very interesting question from the business perspective other than it puts competitors from two separate market segments where they are notin competition with into a new market segment where they are in competition with each other.

PostgreSQL is overkill for most tasks

[brokeninside]

MySQL is all the database that most applications need. Consequently more people use MySQL than PostgreSQL. Most (not all) organizations that need the features of PostgreSQL that MySQL lacks already have Oracle, DB2 or Informix licenses and tend to use those. Not only that but MySQL for a long time was superior to PostgreSQL (lower hardware requirements, faster to configure, higher performance on simple queries when used as a web front end) in very influential markets during the dot com boom. The result is that there are a far larger user base of people with in depth knowledge of MySQL than for PostgreSQL.

Or another way to look at it, for a long time MySQL was the quick and dirty solution. In many IT market segments, the quick and dirty solution is the most attractive solution. Consequently, MySQL is one of the most (if not the most) popular open source database because it was good enough at doing what needed to be done at the right time. From an evolutionary perspective this is all that is needed.

Re:PostgreSQL is overkill for most tasks

[Marcus Green]

Quick and dirty is always dirty and in the long run never as quick.

Re:PostgreSQL is overkill for most tasks

[perkr]

There is no such notion of "always" and "never" in this context.

David Axmark

[Anonymous Coward]

I met one of the co-founders at our Linux user group (www.flux.org). He's a great speaker and *extemely* humble. If you get a chance to see him speak, do so..

Something was said

[Anonymous Coward]

Why is there always some nutjob who has to shit on everything? I didn't see any intricately detailed tech questions, so the poor guy couldn't have answered them. On the other hand, I saw a series of questions which intelligent slashdotters knew would be answered by a CEO, not a coder, and the guy did a very nice job of answering openly and inteligibly.

Go back under your bridge, troll.

That's not true

[brokeninside]

As a bona fide geek, I'd much rather the world run on TRDBMS (where the T stands for Truly) systems rather than the SQL engines being touted as RDBMS solutions by the big database vendors. But the fact of the matter is that few business applications require a rigorous database schema and enforcement of Relational Integrity so that the results of various queries can be theoretically proven to be correct rather than being merely being tested and passed against use cases. Situations that require this level of rigor do exist. People in those situations will tend to use a more robust database system.

But if you're making a web discussion system or a relatively simple retail site, there is not a pressing need for that depth. And in a world of scarce resources, your time is often more valuable being put to other tasks. Frequently, the money meter is ticking down and the choice is between a quick and dirty solution and no solution. In that case the quick and dirty solution is clearly better. And in many cases, there is no need to expand it beyond what it is. Not all problems for which quick and dirty solutions are possible ever require scaling up into something more refind.

Re:That's not true

[plopez]

But if you're making a web discussion system or a relatively simple retail site
WOuld having your credit card misbilled or sensitive information sent to the wrong person change your mind?

Or if your comments are misassociated with another user's and the Department of Homeland security begin to take an interest in you would you change your mind?

In other words, there is nothing trivial about it. 'Good enough' is just a politically correct term for 'slipshod' or 'crap'.

But what about the future?

[Anonymous Coward]

The question was about the future. While you're correct, MySQL may be edging into the commercial market at this time, what's to really stop PostgreSQL, or even SQLite, from eventually doing the same?

I gather than MySQL is only ahead at this point because more people know of it. It does get far more press coverage and shelf space at bookstores than PostgreSQL gets. But people are beginning to learn more about PostgreSQL, and it's use is growing. The same goes for SQLite.

Frankly, I don't think many companies give a damn about contributing to the FSF, fighting software patents, and so forth. What matters is their bottom line. And if going with PostgreSQL will save them licensing costs, due to it being released under the BSD license, then they'll go with PostgreSQL. They'll be even more eager to go with PostgreSQL when they find out it offers enterprise-grade features that MySQL still lacks. There are enough firms and consultants out there who offer reasonably priced support, so that's not really a problem. In the end, the reasons for going with MySQL appear quite minimal, while there is much to be gained by going with PostgreSQL. How will this sort of decision making in favor of PostgreSQL be addressed?

Re:But what about the future?

[martenmickos]

OK, sorry for not understanding the question. Here is what we do about the future:
 
Thanks to its popularity, MySQL has become a platform for innovation. By this I mean that increasingly, other innovators are attaching their leading-edge technology to ours, thus enhancing the overall competitivess of the product.
 
We did it ourselves with MySQL Cluster. That's technology originally developed by Ericsson - and today it is an absolute category leader in telco and networking. Nokia, Alcatel and Nortel are all building real-time network nodes on top of MySQL Cluster. No one else has a main-memory based shared-nothing cluster with that high throughput and availability. Or take the new Monitoring and Advisory Service that we are launching as part of MySQL Enterprise - this is a novel innovatoin built on the feedback from our most advanced users and customers.
 
As for partners, there are companies developing special-purpose storage engines for MySQL, with the help of which we will be able to lead in new markets. Others develop encryption technologies, data synchronisation, or something else. Whether it is produced by the community, by partners or by ourselves, these innovations will help provide the new solutions that the world needs.
 
I think you are saying that software business is essentially an innovation battle, and I agree with you. We innovate in production models, organisational models, business models and most definitely in software technology. But, of course, we must not get complacent.
 
Marten

Where were the real questions?

[anomalous cohort]

I'm surprised that there were no questions about the real crisis facing MySQL which is that their two most reliable transactionally aware storage engines, InnoDB and BerkeleyDB, are now both owned by their competitor Oracle.

Isn't anyone else here concerned with that or is it really true that MySQL is only used as a relationally aware ISAM?

Re:Where were the real questions?

[T-Ranger]

And both of those are under OSS licenses, so if Oracle starts being a dick about them, MySQL AB can just fork'em. Depending on how much of a dick Oracle becomes, AB may be able to hire away entire dev teams and only miss a month or so of working time.

The question boils down to if you think Oracle bought these companies/products in an effort to shut them down, or to continue working on them and making money on them. The former is just absurd as the risk of failure (from Oracles perspective) is almost infinite: if Sleepycat was making money on an OSS product line, and Oracle buys, and kills off Sleepycat, then they diddn't kill the market: someone else can form Wakeycat tomorrow and continue.

Re:Where were the real questions?

[martenmickos]

Great question and I am happy to address it. I'd say there are three main responses to your question:
 
First, you can't kill a GPL product by just acquiring it. If you as the owner continue to maintain it (as Oracle has done so far), nothing bad has happened. If you stop maintaining it, the community will quickly take over and make sure the software stays uptodate.
 
Secondly, I believe that the open source world has a self-healing mechanism built in. When something disappears or is at risk at disappearning, replacements quickly emerge. Completely on his own, Paul McCullagh developed the transactional PBXT storage engine for MySQL. And our partner Solid did the same. Check out their website http://www.solidtech.com./ There are other storage engines in development and under design.
 
Thirdly, as a company we are ensuring that there are strong transactional storage engines available for MySQL. MySQL Cluster is one and the Falcon project is another one. Falcon is a modern transactional storage engine built by Jim Starkey (known as the inventor of MVCC and the one who coined the term blob) on basis of technology he had developed and that we acquired. Falcon will soon come out as alfa.
 
The acquisition by Oracle of InnoDB (i.e. Innobase Oy) sent some shock vibes through the open source community a year ago, and perhaps rightly so, but I think we can be proud of what has transpired since then. As for Sleepycat (and their engine BerkeleyDB), it never was in any real use inside MySQL and after looking at the technology we decided a long time ago not to pursue that alternative.
 
Overall, there is another learning from all of this. It is a step towards increased modularity of MySQL, which in itself makes it easier for others to contribute code. We have taken an "architecture of participation" as a priority for us, and making the product modular is one important step in enabling participation. If you have good ideas on how to achieve higher modularity and enable more contributions, let us know!
 
Marten

That is true now, but it wasn't always

[brokeninside]

For quite some time, installing PostgreSQL on Windows was either a long and complicated process involving manual creation and configuration of user accounts and manual registration of services or installation of Cygwin. Back in the late nineties and early oughts when MySQL started to gain its mindshare ascendency over PostgreSQL, getting PostgreSQL to run well on Windows was something of a tiresome task. For some of us, the payoff was well worth it. For those looking for a quick and dirty solution, it wasn't worth the hassle.

Bi-temporal capabilities?

[nuttycom]

Unfortunately, I missed the initial call for questions, but are there any plans for MySQL to support bi-temporal queries? Tracking history of values in a database is a problem I'm constantly running up against, and there doesn't seem to be a vendor out there who is addressing the problem.

Re:Bi-temporal capabilities?

[Anonymous Coward]

That's a real problem and it's a sad thing that, although we have both the theoretical ground ("Temporal Data & the Relational Model" by Date, Darwen and Lorentzos) and SQL proposals (along with a good free book "Temporal Databases" by Snodgrass), everyone still ignores it. Not a surprise considering almost every database practitioner is really nothing more than an amateur. The relational model is not rocket science, yet 9 practitioner out of 10 have no clue about it (beside babbling buzzwords). Temporal support for databases will introduce huge benefits in almost every scenario, yet people doesn't even have a clue about it and go after other functionalities to improve performance at the physical level. I don't think anyone will introduce temporal support until a lot of customers starts asking for it, and unfortunately I don't think it will happen soon. Just to remind the average understanding of databases let me quote the documentation (dated 2001) of a product ;-)

"Reasons NOT to Use Foreign Keys constraints: There are so many problems with foreign key constraints that we don't know where to start: Foreign key constraints make life very complicated, because the foreign key definitions must be stored in a database and implementing them would destroy the whole "nice approach" of using files that can be moved, copied, and removed. * The speed impact is terrible for INSERT and UPDATE statements, and in this case almost all FOREIGN KEY constraint checks are useless because you usually insert records in the right tables in the right order, anyway. * There is also a need to hold locks on many more tables when updating one table, because the side effects can cascade through the entire database. It's MUCH faster to delete records from one table first and subsequently delete them from the other tables. * You can no longer restore a table by doing a full delete from the table and then restoring all records (from a new source or from a backup). * If you use foreign key constraints you can't dump and restore tables unless you do so in a very specific order. * It's very easy to do "allowed" circular definitions that make the tables impossible to re-create each table with a single create statement, even if the definition works and is usable. * It's very easy to overlook FOREIGN KEY ... ON DELETE rules when one codes an application. It's not unusual that one loses a lot of important information just because a wrong or misused ON DELETE rule. The only nice aspect of FOREIGN KEY is that it gives ODBC and some other client programs the ability to see how a table is connected and to use this to show connection diagrams and to help in building applications."

Postgres vs mysql

[grrrgrrr]

mysql is more difficult to use then postgresql because of missing features and features that do not work as expected. For example when you try to join views that just works on postgress (and other databases) but mysql can not use its indexes so that the join is not realy feasable time wise . Not having views makes a lot of problems more difficult. The stored procedures are nice but also have some terible rough edges that makes using them very difficult. having to think about what kind of table to use is also not userfriendly and i think it is a source of confusion on features and benchmarks (yes it does support transactions and yes it is fast but not at the same time. inno db is very slow!!! compared to the competition )

Re:Postgres vs mysql

[Toba82]

Could you be more specific? I'm MySQL 5 in my current work and if there is a problem with lack of index utilization when using views I could see some serious problems down the line.

More to the point, what version of MySQL have you experienced this problem in?

Re:Postgres vs mysql

[grrrgrrr]

it was version 5.0 just after the final realease maybe 5.1 is better?

Re:Postgres vs mysql

[Anonymous Coward]

You are talking out of your arse. InnoDB is extremely fast when put head to head with Oracle and PostgreSQL. We have an application that must run on all three databases, MySQL rips the other two to sheds when generating reports consisting of thousands of queries.

Re:I've never been asked about PostgreSQL or SQLit

[dskoll]

You've never been interviewed at Roaring Penguin, I guess.

There are tons of companies using PostgreSQL, but a lot of them tend
to be quiet about it. Quite a few "black box" appliances use Pgsql under
the hood.

Problem with different table types.

[The_Spud]

I have seen statements where people from mysql have said that the pluggable storage engines for mysql ( e.g. MyISam , InnoDB) are a big advantage. I have found the differing storage engines to be troublesome in real world use as you have to take into account the table type to know what features you can rely on. For example

no referential integrity/ transactions in MyISam

no full text searching in InnoDB

Differing backup and restore procedures for the different engines.

Are there any plans to upgrade one of storage engines to have a full feature set so that you could use that by default then use one of the others if you require a limited feature set but high performance?

Re:Problem with different table types.

[martenmickos]

I should leave the technical questions to our technical experts.
 
But quickly, the benefits of the storage engine architecture is very tangible for advanced users with heavy loads and different loads on different sets of data. Over time, we will work on streamlining the api's and the syntax so that you don't have to be concerned about this issue. But, typically, those who need to use multiple storage engines typically also will be happy to make use of their specific features although they slightly differ from each other.
 
Marten

Re:Problem with different table types.

[Jamesday]

We're adding many of those features (fulltext searching, common backup) to a layer above the storage engines, so all engines will eventually have many of them. One particularly important project is a new backup system that makes it easier to handle all of the engines. MySQL Forge has the design documents for the new streaming online backup API and feedback to the team is much appreciated if you see any ways to improve it.

Transaction support needs to be in the storage engine and it's useful to have storage engines that are optimised for non-transactional storage. Whether a particular table needs transaction support is a choice for the application developer, not the database developers, since it's the application developer who knows what the application needs.

James Day, Support Engineer, MySQL AB.

Re:Problem with different table types.

[amanda-backup]

> Differing backup and restore procedures for the different engines.

Actually Zmanda Recovery Manager for MySQL provides a uniform method of backing up all storage engines.

MyISAM is my friend

[pestie]

Many of the people who trash MySQL, and MyISAM in particular, seem to come from a formal CS educational background, or at the very least were trained on commercial RDBMS's that supported all those whiz-bang features that MySQL lacked for so long (and still lacks, in a few cases). I, on the other hand, knew next to nothing about databases when all of a sudden I had to build an entire infrastructure around one, with next to no money. Yeah, I know, this is insane, blah blah blah, but that was 8 years ago, the company itself is now 10 years old, and the gamble paid off. My point, however, is that I didn't have these preconceived notions about what a database server was supposed to be, so I just worked with the limitations of early versions of MySQL and grew as its feature set grew.

Even today, I find MyISAM extremely useful. Many of our data sets are big (hundreds of millions, or well over a billion, records in tens to hundreds of gigabytes) but are essentially read-only. For these, MyISAM is great. I can load and index the data on a machine with tons of RAM and fast disks, then copy it to much less powerful production machines and, with all the heavy lifting done, I have lightning-fast access to my data. It's great. Backups and crash recovery are much easier when you can just resynchronize sets of files using rsync, never touching all those huge tables full of read-only data.

ACID compliance and transactions just aren't necessary for everything a database server does in all contexts. Sometimes the reduced overhead and simplicity of the problem means MyISAM is the best choice.

fork 'em?

[kpharmer]

> MySQL AB can just fork'em

right, because transactional engines are:
  - because they are trivial
  - because the need for consistency with the rest of the storage engines isn't important
  - because developers are a dime a dozen
  - because healthy and productive open source projects are such a breeze to create
  - because none of the existing developers in the employ of oracle are encumbered by ip agreements
and most importantly:
  - because

Re:I've never been asked about PostgreSQL or SQLit

[Anonymous Coward]

Just do a search on Monster. 101 job hits come up on Postgres. 3 job hits come up on SQLite. Thousands of hits come up when searching for MySQL.

Thousands of hits come up when searching for Microsoft Access, too. That does mean I'd want a job where they were stupid enough to use it for anything serious.

When will get MySQL get full blown sequences

[Anonymous Coward]

Martin, here's hoping you are reading at score 0.

Every time I see a MySQL presentation, I ask the same question.

When will MySQL get sequence support.

The answer is generally the same. A shrug, something about how MySQL already has AUTO_INCREMENT, and "why would you need anything other than that?". Maybe you don't have the best speakers in Australia, but they've seemed woefully ignorant of the entire point of what a sequence does.

For a number of more interesting algorithms, sequences are amazingly useful or utterly essential. Things like data structures with NOT NULL circular or self references.

Things like the ability to take a whole bunch of data out of a database (complete with references) and clone it (by pulling all the ids in advance and remapping the ids to the new set, then inserting them all at once) or to "cut and paste" large chunks of data with complex schemas into a different database. Or having shared identifier spaces (by having more than one table use the same sequence, etc). The list of things you can't do without seqeunces goes on and on...

Workaround for these in the MySQL case has always seemed to be that you need to insert things correctly in the right order, based on the individual application (which with the circular case above isn't possible of course). And yet with proper sequence support, you can do all these things on arbitrarily complex schemas with a single algorithm/library and not have to continuously reimplement the (often hideously complex) functionality over and over again.

So my question to you is the same as to the others, since I'm assuming you at least understand what a sequence is and what it does.

When will MySQL finally get sequence support? If it won't ever get it, why not?

Re:When will get MySQL get full blown sequences

[Anonymous Coward]

Damn I wish MySQL supported sequences, for the reasons you state.

Re:When will get MySQL get full blown sequences

[martenmickos]

I am reading you.
 
Thanks for the proposal on sequence support. In our company I am not the one to decide on features or present timetables or roadmaps on them. But I am very happy to take your suggestion to our engineers. We get tons of suggestions for new features, and it is difficult to decide which ones to do and in what order. The more people want a specific feature, the more likely we are to implement it.
 
And perhaps you could help us? If you can design it for us, or even contribute code, that goes a long way.
 
Marten

Re:When will get MySQL get full blown sequences

[RobinSchumacher]

Hi - Marten passed your request onto me and I'd like to thank you for letting me know about your need for sequences in MySQL. Like you, I like sequences and used them a lot in my Oracle days so I know how valuable they can be. To be honest, I haven't fielded a lot of requests for sequences, but that doesn't mean we can't look into adding support for them. We're about to launch a new product feature survey (likely in early Nov on our Dev/Community zone), and I've now added sequence support to the poll because of your request. Hopefully, we'll get a feel for how many others like you are out there in need of sequences. If you or anyone else out there has more input on this subject, please shoot me a mail and let me know. Thanks again for the request. --Robin rschumacher@mysql.com

Re:When will get MySQL get full blown sequences

[Omkar]

I don't use any database software, but I just wanted to tell you how impressed I am by this interview and your responses in the discussion. As a student who is trying to decide on a career path, your example is heartening - most businesspeople I have met have been vaguely slimy.

Re:When will get MySQL get full blown sequences

[martenmickos]

Thanks for the kind words! It is always stimulating to hear from others what they think of us. (Of course it is most stimulating when it is positive, as in your case, but if not stimulating it is at least useful when it is negative.)
 
Marten

Database innovations

[jd]

There have been many attempts to distribute databases, but there are enormous problems with the approaches typically used (point-to-point makes for very slow synchronization, nobody seems entirely sure what/where/how to parallelize, there are usually single points of failure, etc).

Another area that has been worked on has been to allow database developers to embed more and more complex SQL-based scriptlets and "helper" functions into the database. However, it is a truism that interpreters (even of bytecode) are painfully slow and offer nothing that a module plugin mechanism wouldn't have (provided you could install modules via SQL statements) and most developers would be better leaving data processing to the clients rather than the server anyway. (The closest I've seen was Informix' blade technology, and that was horribly unreliable and tedious.)

The bottom line is this: There have been a lot of database innovations over the years. Some - I'd say most - have been great ideas but the popular interpretation and implementation has been terrible. This is not the fault of any of the ideas, but I would take it as suggestive that programmers are not being as imaginative or creative as they could be.

My question? How can a database company:

• Combine the artistry vital to any creative effort with the architectural precision needed to make the art something special?
  
• Keep things recognizable enough to be usable, without repeating everyone else's mistakes?
  

That's advice for another database

[Anonymous Coward]

Specifically it is common (and good) advice for Oracle.

There is overhead to running a query optimizer. MySQL chooses to have a shitty optimizer which takes very little effort to run. It is therefore not a big deal to not hit cache very often but the flip side is that MySQL tends to choose poorly when it needs to generate a complex query plan. Oracle makes the reverse decision, it has a good optimizer but the optimizer runs slowly. Furthermore there is a global lock involving that optimizer. The result is that a very good Oracle database can fall over if you're preparing too many different queries. Instead prepare just a few queries and use placeholders. That way you can hit cache all of the time.

So it is good advice. It is just good advice for a different database. :-)

Re:I've never been asked about PostgreSQL or SQLit

[grazzy]

.. nor have we ever heard about it.

Also, MySQL is more common on web-hosters

[walterbyrd]

On the lower priced web-hosters, PostgreSQL has been almost non-existant. That is just begining to change.

Apparently, you're not out of school yet

[brokeninside]

In the real world things come in degrees. What constitutes ``good enough'' changes with the industry segment you're in. The same applies to just about everything. What is ``good enough'' for a folding card table in a frat house isn't the same as ``good enough'' for a formal dining table for a governor's residence. And there are many gradations in between.

This doesn't necessarily mean that it's slipshod or crap. It means that a conscious descision was made as to what level of qualify needs to be enforced and testing was calibrated to meet that level given the number of available resources in time and money.

How do you pronounce Mårten?

[mostlycurious]

Hi Mårten, I can't pronounce anything that's not on my keyboard, so how do you pronounce your name?

Re:How do you pronounce Mårten?

[martenmickos]

I pronounce my name in whatever language I am speaking. So in English, it is something close to "Martin". In my native Swedish language, it is pronounced like "Morten" with a rolling R. You may have noticed that in Norway, that's how they spell that name too ("Morten"). And in Denmark they used to write "Maarten", which at the same time is the most official way of transcribing my name. But I like simple solutions, so feel free to spell it Mårten or Marten and feel free to pronounce it any way convenient for you.
 
Marten

Re:How do you pronounce Mårten?

[mustafap]

This has been the most inspiring collection of responses from any CEO through any media I have ever come across. It will certainly make me think twice about about my database product recommendations next time round.

Thanks for spending the time to listen to us geeks :o)

... because SQLite and PostgreSQL just works.

[Augur]

Tryed to do in MySQL nested queries with LIMIT in inner circle?

Why MySQL up to 9x times slowers in trivial tasks (selects) compared to SQLite?

I have 400k rows of heavily crunched data (telecom CDRs) and SQLite just works, faster and painless.

n00b!!

[anomaly]

Who let new people in the pool? :)