How is a company that gives everything away for free worth $1bn? The only way to make money is from support. That will come as a shock to some people who haven't dealt with the 'free software' lot before.

User:

'Hi, I downloaded your free database and now I have a problem'

Sun:

Sure, we'd be happy to help, but it'll cost you $1000 to speak to support for 15 mins

This absolutely fits in with the direction much of Sun's software has been moving: the software is free, but you can buy support for money. I can well imagine that Sun will make this acquisition profitable by offering:

Paid for support

Training

Publications (cookbooks, etc)

At the same time, the existing MySQL on Lintel customer base may get to see huge performance improvements from MySQL on Solaris/Niagara. And MySQL on Solaris/ZFS on a X4500 could become the definitive poor man's data warehouse. And then you imagine a bit further ahead: MySQL on a stripped back Solaris base as a SQL appliance (following a similar train of thought to the Nexenta storage appliance).

Of course, we've seen Sun acquisitions which didn't take - Cobalt springs to mind - but this one brings a large new potential customer base. It also gives impetus to the Solaris x86 OEM deals with IBM, Dell et al - will customers want to run Sun MySQL on RedHat Linux on HP server, or Sun MySQL on Sun Solaris on IBM server, or Sun MySQL on Sun Solaris on Sun server?

I look forward to hearing Ashlee and friends chewing this over on the next Open Season.

Does that mean they'll be charging for it, too? Presumably if they're paying a not unsubstantial sum for it, they'll be reluctant to keep it free, won't they?

How does this fit in with Oracle owning InnoDB (one of the table types available in MySQL)? Is MySQL going to get pulled in two different directions now, or abandon InnoDB in favour of Falcon?

... so Sun must think that it can make even more people want to pay for it. Being backed by Sun, gives mySQL the backing/support of a large corporate that a lot of companies want to see, before they will commit to it - If they'll commit to it, they'll want it supporting = $$$ for Sun where they would have gone elsewhere i.e. Oracle, Sybase (Yes, I'm aware mySQL isn't in the same league as those two)

MySQL is open source, so will, in some form, remain free in both senses.

It ain't free. Well, not completely.

It's free to download and free to use if you're not deploying it commercially.

Otherwise they do a range of webserver and CPU licenses, costing a few hundred quid upwards. Not sure how adept they are at chasing up unlicensed users mind you.

It's GPL, so someone could branch it and keep developing in the unlikely event that Sun closed the source.

Anyway, people could just switch to the far-superior PostgreSQL and the world would be a better place.

It's not just about support. MySQL give away the basic package but have a number of enterprise products they'll happily sell you as well as support and training etc.

This is excellent news: 1. it keeps MySQL out of the clutches of Oracle (who would certainly just kill it) 2. maybe Sun will develop better multi-threaded support (e.g. non-blocking slave replication updates) to "leverage" the Niagra chips etc. 3. gives Sun more incentive to optimise Solaris x86 to displace Linux (not that that would be a good thing, but improving Sol86 further would do no harm)

For those concerned how MySQL could be worth $1bn when it "gives its products away for free"..

You are overlooking the fact that MySQL is available under a dual licence. Those who incorporate MySQL into a product distributed under a non-free licence must pay for the Enterprise edition of MySQL. Similarly those who want support and a more traditional relationship with their database supplier (MySQL) can pay for the Enterprise Edition.

If you do not distribute your product or distribute it under the GPL then you can use the free (in both senses of the word) version of MySQL.

Haven't looked into the valuation or anything, but the move itself sounds like good strategy.

Hope it's executed well. :)

First off, all the major vendors offer free versions of their database products and then make money on the grown up versions and support. MySQL is no different in this respect.

It'll be interesting to see how Sun spin this with regard to Oracle as they really don't want to upset Larry and lose all those high-end server sales that Oracle drives.

I suspect they will position MySQL in the lower segment well away from having any impact on Oracle DB sales,

I use MySQL quite a lot and hope Sun look after it properly, especially keeping a reasonable parity with the dual licenses in terms of functionality and fixes.

I don't think Oracle will lose too much sleep as MySQL isn't really a competitor once you get a reasonably high transaction rate. On the other hand a lot of companies need less database power than they think!

This makes no real difference to me as a MySQL user\implementor ... I can download it and use it for free and where I've put it in a client site and need some support then I can pay for it.

They seem to have paid way over the odds but that ain't a first.

MySQL AB made money from the traditional FOSS methods. Support, training etc.

They also came up with a novel way of making money from their development cycle.

If you wanted something that they did not offer then you could wait for them to develop it, or pay them to develop it for you. Several features made their way into the MySQL code base because someone was prepared to fork over the cash for a developer.

Yes, it is, if you want to follow the GPL:

http://www.mysql.com/company/legal/licensing/opensource-license.html

imagine a 64bit mysql backend running on a 64cpu sunfire box with a 2tb distributed mysql cluster RAM cache up front serving the indexes... wire it up with low latency dolphin sci interconnects and an optimise the whole shebang to run on solaris....

then imagine a cluster of those!!!!!

As others have stated MySQL is not free for corporate use. Having it in the LAMP stack is a fantastic way of crowbarring their way in.

Case in point: In my last job, another team used it as a backend for customised x86 boxes that went into shops, a few thousand of them. We sold these for a couple of grand. When I realised they were using MySQL I pointed out the licensing costs, management panicked, and phoned them up for a quote.

As each box was deemed a server, we got quoted around £600 *each*, a quarter of the whole unit price. This got knocked down substantially, but not enough. So we simply reengineered it to use sqlite, which was a better fit anyway (we didn't do anything fancy, just plain SQL)

I imagine there is a large seam of gold waiting should Sun go hunting for unpaid MySQL installations.

Sun now has a now bought all the licenses they need for the low-end side of their utility computing model.

I really don't think it will be much longer (3, 4 years) before may small to mid-size web projects simply lease DB space from companies like Sun and connect over the internet -- or just across the colo.

For me, it would mean not managing drivespace for backups, when is it time to buy another server, how to balance RAM between different applications, etc. Just rent capacity from high-performance, optimized database servers.

The developers at work, on their side plate, already have one project this year of moving much of our routine image storage and resultant bandwidth off to Amazon S3.

And MySQL has a huge role to fill without becoming Postgres or Oracle or Sybase -- and my background is largely MS SQL & Sybase (both before and after the split), and I'm doing some Postgres tuning this week. But for the last year or two I've done a lot of MySQL. So I've played with many of the rdms's and find they all have advantages and disadvantages.

For databases that need more flexibility then SQLite, but aren't that sophisticated to need the features of Postgres...I find MySQL easier to setup and tune then Postgres.

For anyone with a ten or twenty million and would like to start a database utility company, fell free to hire me :)

A standard full install of Solaris 10 already includes Mysql, Postgres and Sqlite. I wonder if Richard is going to get an offer that is impossible to refuse?

Can this deal get us check constraints and selects in triggers? Because if so I would be a much happier DBA.

According to this rumor mill site http://www.theregister.co.uk/2005/11/17/sun_solaris_postgres/ Sun has already committed to PostGres.

Sounds a lot like the old Sun litany of "Solaris on SPARC and nothing else"! So, you're suggesting Sun concentrates on pushing MySQL on Solaris and upsets the main MySQL customer-base, namely Linux users? The purchase only makes sense if Sun keeps making money out of MySQL, and to do so they have to keep the Linux users buying it because the MySQL on Solaris 10 on anything market is just not big enough by itself.

Oracle will probably feel safe in the enterprise space, but with M$ SQL eating them from the low-end upwards they'll probably not be too happy with Sun giving more muscle to another low-end competitor. It will be interesting to see Oracle's response, especially given Larry's intentions on the Linux market. M$ will probably be smiling - they probably view Sun's embrace of MySQL as the kiss of death to what was becoming an irritation in the low-end space.

Mind you, this is a much gutsier move than the StorageTek fiasco, and done right (i.e., without annoying the other hardware vendors that provide 99% of the servers that commercial MySQL is run on, or the Linux distributors) it could actually make Sun some money. Hopefully enough for the move to Itanium when they abandon SPARC/Rock/Niagara! ;)

I'll have the jacket with the "Schwartz lied and SPARC died!" logo, please....

Stop it, I'm at work, and you'll have me making unprofessional faces (and noises) if you continue!

I'm a bit jittery about this.

I used to own a couple of Cobalt RaQ 3 servers a number of years ago and the purchase of Cobalt Networks by Sun for a £4 billion was seen as a good thing. They were going to develop the app code and build the product line. After all those pretty little blue box's were piled high in data centre's that world over.

Problem was that pretty much as soon as Cobalt was in Suns hands the software updates stopped dead and development of the application code died. Sun released a couple more servers under the RaQ name and that was it. It didn't take long for the market to move to blade servers and the whole Cobalt line up was EOL and buried, gone forever. They did open source the code which has morphed into BlueQuartz but that's hardly the greatest achievement possible with a billion dollar+ investment.

Now I know that this situation is different. I hope that Sun need MySQL more than they needed Cobalt and that I'm worrying about the sort of thing that Oracle would do to MySQL not Sun, but...

Hopefully I'm fretting over nothing.

The owners of MySQL wanted to sell their well-marketed piece of shit anyway, Sun needs customers and users will happily continue to use a piece of poo. Plus ça change...

Meanwhile Sun is also getting into bed with Postgres (which it can't buy)cf. Postgresql.org news-

You can ask any crack dealer about the merits of giving away some of your product.

It builds a loyal customer base, who are willing to pay for more.

"You can ask any crack dealer about the merits of giving away some of your product.

It builds a loyal customer base, who are willing to pay for more."

And aint that the truth, Simpson. In the Binary Sector though, the Virtualised Zone which Digitises Information to and for Intelligence so that IT and its Media partners/customers can Present that Intelligently Designed Signalling to Sight and Sound showing us a Reality based upon Coordinating Instruction Sets Commandeering Output, [which may also be Originality rather than just a Cloning of Old Codes], New Code[s] in Original Content will always lead with a New Present Picture/Future View which is supportable/supported.

Whenever you own the Infrastructure to Create a Virtual Picture for Real from NeuReal Content, can you Create the Future in AI Beta Program and ITs Beta Programming, which other Systems and their Administrations/Attendant Structures will have to Better if they are not to Follow. And whenever you are not "owned" and Servering to Biased Instruction Sets/Covert Sects/Big Business Big Brother Interests as is the Beauty of Raw Open Source in Transparent Code for Readily Available Third Party Adoption rather than Sale/Licensing ...... because of the Inherent Added Mutual Benefit to All Parties by Virtue of the Following/ITs Code Adoption .... will that Virtual Picture become a Shared Reality Created by Advanced Artificial Intelligence Miens/Memes [which can Work Rest and Play as Alien] with a Physical Mirror of the Virtualised Landscape down on Earth.

Such is CyberIntelAIgent HyperRadioProActivity ...... AI Prior Art Discipline with ITs Patent Rights Pending and Reserved in Copyright even though they be Shared Copyleft for Adoption as a Virtual AIdDevelopment for Paradigm Change enabling Future Construction for Present Structures.

I'm not a database expert and therefore when I need to use one I prefer it be as "correct" in its implementation as possible. It's not that I'm going to hammer it, but I prefer that if there are faults or problems with SQL that I can lay the blame squarely on my programming and not some esoteric behaviour of the database. I also want the database to be strict and slap me if I write bad queries.

I have had very good experiences with PostgreSQL, mainly on Windows. It has an MSI installer plus all the bits and bobs you need to make it work with ODBC, .NET, Hibernate, excellent help, admin tools and so on, all in a small download. What I like is that it installs straight out of the box and the administration tool called pgAdminIII allows me to drill down and execute arbitrary queries. The tool used to be a bit crashy in earlier versions but these days it's very stable. I've also gotten PostgreSQL working with OpenOffice Base for form entry. It also works great with Hibernate with JDBC and that's what most of my coding is through.

I'm sure MySQL and Postgres advocates can argument about speed or configurability until the cows come home, but for my money Postgres just feels right and correct which is more important to me than speed.

comes from licensing, afaik

Anyway I suspect this will herald in Postgres's ascendancy

Referring to Enterprise Server: "MySQL Enterprise is the comprehensive subscription offering that provides database developers and DBAs with everything they need to successfully develop and deploy database solutions with MySQL. It includes Certified Software, updates and upgrades, proactive alerts and advisors, the online MySQL Knowledge Base, and full production-level technical support. The Certified Software (database server, connectors) is provided under the GPL License. Optionally, customers may choose a Commercial License."

That is from http://www.mysql.com/company/legal/licensing/

So comments that one cannot use MySQL server under the GPL for commercial use are unfounded, for now...

Postgres. MySQL has been good to me for years, and I trust that someone will fork a GPLed "new, new version" if Sun somehow can't quite bring this through their Reality Distortion Layer without catastrophic damage, but I'm going to make darn well sure that anything I do on MySQL from here on out is trow-a-switch portable to Postgres and/or Firebird for the next year or two. Which should be a standard practice anyway, no?