Sun's MySQL fork survival theory ripped If Oracle screws up MySQL, the community will fork and the database will live on under another name - leaving Oracle high and dry. At least that's the open-source theory. And it's a theory Sun Microsystems' executives past and present have recited to placate those concerned by the prospect Oracle, the number-one database …
The comments about customers switching from Oracle to MySQL speak of a complete lack of data and/or knowledge of the market. My old company surveyed a large number of people switching from Oracle a couple of years ago, and, if they were switching, it was to Postgres, not MySQL. In fact, we found more people switching from MySQL to Oracle than any other movement between proprietary and open source.
Mueller may have been an 'adviser' to MySQL, but I've done open source strategy for a huge number of companies in the last eight years, from startups to Global 50's. None of them have ever been interested in moving from Oracle to MySQL. Oracle to Postgres, maybe, esp. with EnterpriseDB's Oracle emulation, but even that was a stretch. Most people will not switch from Oracle because the risk is too high and there is not enough MySQL talent available to mitigate that risk or maintain production systems.
Takeshi Tachi of NTT is on the record saying they are switching from Oracle to PostgreSQL for some mission-critical apps and they expect to save $10M a year. PostgreSQL is not owned by a corporation as MySQL was, and if Oracle kills, starves or smothers MySQL, they will only clear up space for a far more technically capable competitor product, albeit one until now not as well marketed as MySQL.
Maybe the only way MySQL could survive as a fork would be to make it non-commercial - ie - have a group like the Apache* people take it on (obviously under a new name). If the group was high-profile enough (like the Apache* people), I think it would gain repect pretty quickly.
(*) Other open source software groups are available.
Could you clarify "Many were puzzled by the EU's decision to investigate the deal, given Oracle's known in the RDBMS world and MySQL made its name as an embedded and web database"?
Unless someone changed the definition when I wasn't looking, RDBMS stands for Relational DataBase Management System. MySQL is as much an RDBMS as Oracle.
Although one can easily say, "Mysql is whatever I get at mysql.com", people started using mysql because they liked its feature set. Admittedly some people might get confused by a fork, but most people will be looking for the mysql they know and want and will find it whether it's called mysql, percona or whatever.
He's missing the point, surely. In the scenario we are considering, the MySQL brand would have been nobbled by Oracle. Therefore, any fork would face no competition and have gazillions of eager customers waiting to lay their hands on an already existing product. Even Gerald Ratner could do the brand management in *that* situation.
To their credit, Oracle appear to have figured this out, even if our branding expert hasn't.
Back in 2004, the X Windows implementation used by Linux was XFree86, the XFree86 project changed the license and this upset people in the community. The code was forked and taken on by a foundation called X.org under the previous license conditions, within about 6 months all the major distro's had migrated to back X.org instead of XFree86, they still use X.org now, XFree86 is pretty much non-existent.
In order to migrate from MySQL to a fork, it just needs the support of the major distro's on Linux and a tiny bit of marketing for the other platforms, given that none of the distro's will want to be particularly reliant on Oracle (least of all RedHat), you can be fairly sure that they will get together with the community to back a fork as soon as Oracle takes control.
In the TWiki / Foswiki split, which happened a while ago, both branches appear to be popular now: Sourceforge ranks 346 (TWiki) and 407 (Foswiki) are pretty similar! I haven't seen a recent comparison of how the two projects compare these days, but both look pretty healthy! So perhaps there is room for MySQL to fork quite happily, one branch for Oracle's target market, and the other for … er, somebody else ;)
...only when the open-source guys are promoting it. Forking,allegedly, is one of the fundamental rights of open source software, and is as much as buzzword as copyleft. Yet when a company (read: evil money-fuelled bunch of zealots driven only by profit and greed) suggests it as a solution, it suddenly becomes blindingly obvious to those within the 'free software echelon' that they didn't really think this through and apparently, to them, that Spongebob Squarepants holds more water in this particular argument.
So it appears that the 'with GPL, we can always fork, because it protects people's investments' is about as much use as the police serving your gran with an ASBO, because she told the local scrotes that they couldn't find their own backsides with both hands.
It's not that, in principle, I disagree with the furore that's thrown up because of Oracle buying up just about everything that goes near a computer, but seriously boys a balanced argument is one that can be traversed both ways..... not thrown out with the dishwater when one of your poster children could be adversely affected, by your own rules of thumb
MYSQL doesnt strictly fit under the definition of an RDBMS because it lacks features such as referential constraints...
-------------
Depends which engine you're running it under - true, MyISAM is NOT relational but InnoDB (which was already owned by Oracle) IS - pretty much.
I suspect that Mueller said what he said for other reasons. I think that Oracle are not out of the woods, although they may have thinned them out slightly.
A products 'brand' is not just it's name. You can easily change the name of the product and the product's reputation, performance, cost, placement remain. Witness Marathon > Snicker and Cloudscape > Derby. Both are well developed products that have not suffered simply because names have changed.
Also, remember who would drive the fight back. If the forked db were to be called DATABASE_637$, and maintained the current feature set, the techies who use it would not be fooled that the name had changed. They would see through the marketing. They saw through Vista's marketing to destroy that product, and everyone likes GIMP no matter how it sounds eh?
CKM writes:
"Mueller may have been an 'adviser' to MySQL, but I've done open source strategy for a huge number of companies in the last eight years, from startups to Global 50's."
Hats off to you, then. I take it that you're aware of Florian Müller's reputation - he isn't oblivious to "open source strategy", you know.
And it's not just the "Oracle market", it's the entire database market - that's why people are petitioning the competition authorities. Once upon a time, conventional wisdom stated that if your database system couldn't handle the volumes, you'd take the hit and run Oracle. Some very large volume applications and businesses now run MySQL (and, indeed, PostgreSQL). Going on about foreign key constraints (and presumably MySQL versions earlier than 5.0 at the very latest) is just squabbling over the details.
"None of them have ever been interested in moving from Oracle to MySQL."
If they're start-ups you're talking about, I'm sure the Paul Graham school of venture capital would have something to say about start-ups running Oracle. It wouldn't be flattering.
"MySQL is not nearly where it could be,"
And it never was. Perhaps if the original owners had not continuously sung it up into the ranks of its betters (and almost anything was better once one had actually attempted to deploy this half-baked wannabe tech in the enterprise) they wouldn't be facing such a storm of indifference now they want it back, please.
I mean, what did the world *think* Oracle would do with this bloody thing? The OS/FSF crowd needs to admit the original owners Sold Out to the Satan of Databases (who make a fine enterprise-ready, real-word resilient product for all that) and move on with their lives.
As for the free database needy - Postgres. Almost anyone who has tried it never goes back.
well, AC 15:26, I guess you're not a dba. I guess you've never had to clean up the mess that those mere details would have prevented had they been used ("we don't use checks and unique indexes, our application checks all that". yeah right. Heard that how many times). The purpose of a database is (to be a cornerstone of) data management. Where do suppose data integrity fits into that?
Idiot.
"The crux of the argument is that, yes, you can fork a GPL project - such as MySQL - but the fork is not viable. That's because while you can duplicate the code, you cannot easily duplicate the ecosystem around it or the brand itself that have taken years to build."
Tell that to the guys at Xorg, they obviously didn't get that memo
As both Geoff and Goat Jam pointed out - X.org forked XFree86 and became more popular, CVSNT forked CVS and became so popular most people think CVSNT is CVS (at least on windows) and CentOS forks RedHat and is commonly known as the 'cost-free redhat'.
Yes the 'brand' has value separate to the software, and certainly part of the price Oracle paid for Sun was for the MySQL brand, but successful branding is different to competition.
Having the FREEDOM to use and re-use the source code (and any related patents GPLv3) fundamentally breaks any claim of anti-competitiveness - someone can always compete - that they choose not to or they do so with poor marketing does not mean that someone that does put in the effort and markets well should be sanctioned.
Sun/Oracle should be given favour for releasing code under a FREEdom license - not vilified for it.
BlueGreen writes:
"well, AC 15:26, I guess you're not a dba. I guess you've never had to clean up the mess that those mere details would have prevented had they been used ("we don't use checks and unique indexes, our application checks all that". yeah right. Heard that how many times). The purpose of a database is (to be a cornerstone of) data management. Where do suppose data integrity fits into that? Idiot."
Such data integrity features have been in MySQL 5.x for a number of years. I don't use MySQL actively, nor did I advocate not using constraints or other widely accepted relational database features, but it's clear that your self-righteousness and deficient interpretation of what I actually wrote marks *you* out as the idiot.
Time to get off your high horse, I think.
Here's what I was saying...
Müller may not be right about Oracle shops migrating to MySQL, although he claims to have evidence, but he doesn't have to confine his analysis to that particular area: it's about the choice available in the entire market.
You can argue that MySQL is inadequate and that seasoned Oracle shops would never migrate because the short-term pain would be overwhelming, or that PostgreSQL/EnterpriseDB would be more of a competitor to Oracle in those shops, but the market for database products isn't static: the authorities have to consider what new customers have to choose from.
If you're a company going to manage a lot of data (and aren't one of CKM's start-ups who naturally go for Oracle, and who must have found a lucrative niche to fend off that Graham-style critique about blowing huge sums of capital on expensive furniture... and RDBMSs), then it's possible you'd consider MySQL as something you might work with (as Google and others have done for stuff, and I doubt that it's only used for Sergey and Larry's personal car cleaning roster). If Oracle remove MySQL as a viable choice, they have negatively affected the competitive landscape, whether you, I or anyone else would laugh at those people choosing it.
I suppose the title of the original comment says it all: "the Oracle market". But as I wrote, it's not really about "the Oracle market" at all. Thus, the original comment was wide of the mark. Apologies if I came across as harsh towards your counter-critique, but you weren't responding to a dyed-in-the-wool long-time MySQL devotee after all.
> and that seasoned Oracle shops would never migrate because the short-term pain would be overwhelming
I don't think it's the short-term pain at all but long-term risk versus cost. if you're running Oracle then MySQL is an unknown quantity (and one without a very good reputation). The cost of data management is not the cost of one RDBMS program versus another, but the overall cost. Oracle software can be bloody expensive, but if you factor in the hardware and the annual personnel budget to manage it then you realise it's not such a large part. Also, if you are a multi-hundred-million pound company then sod the cost, your data is worth more.
And I'm ignoring the question of whether MySQL will run as nicely on those 16 processor servers as did your now-displaced Oracle.
> and aren't one of CKM's start-ups who naturally go for Oracle
I don't think he was really saying that, but it is a matter of cost/benefit. The high cost of a competent database can pay itself back very rapidly under the right circumstances. It can be worth splashing out on the big stuff because the big stuff can do straight off what might otherwise take a much more expensive programming effort on top (and it can usually do it much more reliably).
> If Oracle remove MySQL as a viable choice, they have negatively affected the competitive landscape
true. I suspect that as someone above suggested, it would be rolled into a high profile project like Apache. It's a good suggestion anyway.
What frustrates me is when people start treating RDBMS's as commodities; equivalent and interchangeable. Beyond using them in the most trivial way, they are not! There is a reason why Oracle/MSSQL is expensive. They can be extremely reliable, handle huge loads, scale well and be very low in bugs compared to almost any other software I've used. *If* you use them to their full extent, the gap between low-end and high-end dbs becomes very evident indeed. But they often aren't so they look like cost rather than value.
So I guess to some extent we were talking at cross purposes and I apologise for my offensive comment; it was entirely out of order.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
