We were chatting with Jeremy Cole today and he brought to my attention last version of MySQL Community Eddition (5.0.51) was released in November 2007 – over 7 months ago. MySQL 5.0.51a and MySQL 5.0.51b security fixes were released but these can’t be considered proper releases.
If we look at the old Kaj’s Announcement we can see there suppose to be 2 yearly binary releases (which are overdue) and 4 predictable yearly source releases, which we have not seen either.
Though, Honestly, I do not care that much any more. I have adapted my ways not to go to dev.mysql.com for MySQL downloads any more but go to mirror.provenscaling.com, where Jeremy Cole promptly publish MySQL Enterprise sources and binaries. We take sources and add spice them up with other community extensions which can’t ever get included in official MySQL tree and use that version instead.
It is worth to mention at the same time there are some improvements when it comes to development community MySQL now moved sources to launchpad – now it makes it easier for you as contributor to create your branch or grab and test someone elses. If you’re just looking for stable, recent, binary MySQL Community release, there is little MySQL offers you.
I found a MySQL hanging bug 2 weeks ago – the hard way, at 4am on a production machine – and was shocked to not be able to find a Community binary with the fix (5.0.54 or above).
I’m now an Enterprise customer for $600.
Do you think this is intentional?
Tom, I think this is pretty much intentional.
I have myself switched to mirror.provenscaling.com for the Enterprise binaries and bought myself an unlimited license of MONyog monitoring tool. The only missing link is support and I am planning to buy that from companies like Percona, Proven Scaling or Pythian.
I think the above mentioned combination would give you better benefits as compared to MySQL Enterprise but at a fraction of the cost. Lastly, you don’t have to worry about MySQL/Sun salespeople trying to up-sell!
How does Proven Scaling get i5 OS builds? Is there a VMware image for it? That is impressive.
I think it is dead not only because lack of release but as the idea also. There is only single patch SHOW PROFILE which makes the Community release different from Enterprise.
Also it is very unclear how it is going to work in 5.1.
That’s very troubling… The only reason I use MySQL is because of the strong community support. I know that any problem I have is only a Google search away, but now MySQL seems to be moving in the opposite direction.
I agree with Geoffrey. So I guess us CE users will simply be waiting for our 5.0’s to die a slow and painful death? Hmmm… I’m just glad I already got complete ISO’s for the latest SuSE x86 and x64’s as well as latest MySQL CE’s for each build.
Knowing the mysql users, there’d somehow be folkds that’d start another group and further CE on their own. (I hope!) Else I’d have to plan for a major shift in a couple of years. 🙁
P.S. huh…?! –> https://launchpad.net/mysql-server
I asked Kaj, and he notes that Sun/MySQL does keep its commitment for 2 binary builds a year, at irregular intervals.
(quote even OKed by Kaj so you may regard this as official 😉
‘MySQL 5.0.51a and MySQL 5.0.51b security fixes were released but these can’t be considered proper releases.’
Why? 🙂 They went through the same process as every other binary package set that is released, they incurred the same overheads etc.
5.0.51a was cloned 11 January 2008 and released a shortish time after (around the end of January iirc). We might not release *exactly 6 months* after the last release, but you should be expecting a new Community binary release around the end of this month as I understand (barring unforeseen circumstances, standard disclaimers, blah blah).
We’re opening up participation more and more with moving to bazaar, launchpad, the preview releases that we are trying to get community input on (do we get that though? not a lot..) on forge (and the worklog) etc. Sure I think there is a way to go with the release policies as well (many people do!), but those issues generally revolve around the community editions getting *more stable* releases, at the cost of having to wait a little longer for some bug fixes.
That doesn’t really stop people from grabbing the source and building their own releases though – you guys are the perfect examples of that. Anybody could do this from the bitkeeper repositories online before, or from launchpad etc. now as well. It’s interesting that the people that have this ability, and openly prefer to do this, are the ones that complain the loudest on the lack of releases. 🙂
I’m not saying this is the ideal (I too personally feel we should release to the community more often), but we are certainly not moving towards closing out the community more, from my perspective (on the inside), we are in fact struggling the opposite way, trying to open things more (though this takes time obviously). We need to work on incorporating more of the community patches as well, but doing that in a controlled manner, whilst mitigating risk and assessing community interest in those features has to be our first priority (imho). Including patches that are potentially unstable or could cause performance problems would cause far more community outcry – and yes, even the SHOW PROFILE patch, the first large one that we integrated, took quite a bit of reworking as I understand – check out all the stuff we did to clean up the patch as well in bug 24795 before it was even implemented. That was by no means an easy process on our side.
Having the branches out there on launchpad (call them a fork if you want!), is a perfect platform for assessing these community enhancements if you ask me, and I for one hope that a) the ‘community developers’ pick up on this, and uses that resource as a platform to raise community testing and participation of their own patches and b) that ‘we’ carefully watch the most interesting new developments, promote their usage, provide our own feedback to the developers, and try to gather some feedback from community users on each of them. This perhaps does not help with some of the smaller patches that community people have released to us however, and I don’t have a magic bullet for that to offer either (other than more resources directed towards community patch reviews, mentoring, and streamlining the contributor licensing).
Also, perhaps I’m missing something, but how can you proclaim that something that gets downloaded ~50,000 or more times a day, as ‘dead’? 😉
I’m going to state that these are all my own opinions, and not those of my company, right about now. 😉 I’m biased towards believing in the product and people that work on it! 🙂
OK Mark, I’m not on the inside. But “struggling to open up” has just never made sense to me. It tells me that something is just organized towards closed-ness at the deepest level and that there are powerful interests and people forcing closed-ness. Opening up ought to be easier than keeping it closed. Struggling to be open should be like struggling to oversleep.
That’s not what I was trying to convey really – the struggle is not having a bunch of powerful people pushing us towards closed source, because what is happening *there* is quite the opposite in fact:
http://radar.oreilly.com/archives/2008/04/missed-twitter-questions-jonathan-schwartz-web2expo.html
‘… we have no plans whatever of “hiding the ball,” of keeping any technology from the community. Everything Sun delivers will be freely available, via a free and open license (either GPL, LGPL or Mozilla/CDDL), to the community.
Everything.
No exception.’
Can’t get more powerful than our CEO, when it comes to this. 🙂
The struggle (imho) is the many internal opinions on *how* we should open up our development processes, and pushing those through. Just migrating from BitKeeper for example took a large amount of time purely down to the opposing views within the company/development group(s) on which is the best revision control system to move towards for example. There are a lot of bright minds within MySQL, and a lot of personal opinions on how things should get done that cause great debate. 🙂
It’s the debates and pushing things to fruition, that *I* feel is the struggle. Again, just my own opinions..
To follow up also from my side:
* Sun/MySQL has promised two yearly binary releases of MySQL 5.0 Community Server
* Sun/MySQL has not promised any specific timing for those releases
* Sun/MySQL stays true to the given promises
* Sun/MySQL is scheduling the next MySQL Community Server release, to be labelled 5.0.67 and to appear late July or early August
So the rumour of MySQL Community Servers death is somewhat exaggerated, as the saying goes (also pointing to Mark Leith’s good comments).
And yes, the “old Kaj” is getting older by the day.
Kaj,
What is about SOURCE releases. Is the goal of 4 source releases per year is dropped.
PS. Note question mark in the post, I’m not stating I’m asking 🙂
Also the four yearly source releases goal is on. We may need to improve on their predictability, though. Exclamation mark!
Mark,
By Releases I, and I bet most of community members would mean the release of the fresh version with all bug fixes and improvements and not only couple of security touches. Using marketing talk you can update help files only do the new build and QA and call it release.
Regarding launchpad – I have written about it. This is good tool for contributors but community is 99.9% of people who never wrote a single patch for MySQL. This community was passively contributing by testing and bug reports and what they need is just stable recent release in binary form as I stated.
Good to hear the new community release is coming in July though.
Mark – Jeremy does not build all these packages, just redistributes them.
MySQL Server is happily still licensed as GPL which allows redistribution. Though I would think it would apply to source only.
Honestly I think this is great MySQL is not preventing Jeremy from doing this. Though I guess “Obfuscation” is main game here. If you get to web site can’t easily find latest version with bug fixes you will go ahead and buy MySQL Enterprise as Tom states in the first comment.
Thing we should not forget – Sun is commercial company and MySQL group has to make money. Though in my opinion MySQL stretches its accelerated growth practices beyond “ethical open source”.
Why do you count 7 months for 5.0.x version? How about 4.1.x or 3.23.x? Please note that version 5.1.x updates regularly and there is no need for panic. I think 5.0.x is getting old and mysql developers are working on 5.1.x and 6.x.
Dmitry,
MySQL 5.0 is the latest stable “GA” release (5.1 is still RC) this is why I focus on this version. It is also most broadly used MySQL version at this point.
Also note unlike with 4.0 for example there were MySQL 5.0 release coming out during these months. There is MySQL 5.0.62 however there all were customer only releases.
Maybe need an offical group of people who publish mysql enterpries similar to CentOS / RHEL from time to time.
Howa,
The thing with MySQL – Enterprise Version is fully GPL unlike with RedHat which has some non GPL components and so can’t be fully distributed – this is why CentOS is required.
With MySQL you can just get the same as official binaries from third party – http://mirror.provenscaling.com/
Well, I’ve give a try to Provenscaling Yum Repository.
5.0.54 works fine, so I trun auto update on.
But 5.0.60 broken our internal server.
Then I dicide just stick with old-good 5.0.51a.
Ben,
This brings another point. The Enterprise versions are only available (and tested) by handful of enterprise users (compared to millions of community users) and also many of these enterprise users are not rushing to upgrade to latest and greatest 5.0.62 This means MySQL Enterprise versions are lesser tested and so more prone to uncaught regressions than community versions were before.
And community testing effort is now at large extent wasted – MySQL can little gain from 5.0.51 bug reports because many of them are fixed in recent releases (with new ones added)
Peter,
I wish this (limited testing by the community of Enterprise releases) were not so. There are several bug fixes in versions 5.0.54+ that I want and I prefer to use a version that has been widely tested by the community.
Mark,
Thanks for confirming. Hope your voice (as voice of big Reference MySQL Customer) will be heard. Interesting enough I know bunch of other Enterprise customers who were bitten off by Enterprise builds quality so they are in the serious trouble – Community build available is very old and missing many of bug fixes they want and Enterprise builds fix them but they are not tested so much and who knows what kind of bugs they add 🙂
Yes….. there’s NO way I’m going to be in a hurry to test out an ‘enterprise’ binary that hasn’t been tested by the community.
It’s just not going to happen…
This is a GREAT way to get me to avoid ever giving money to MySQL.
Kaj,
I don’t understand how the four quarterly releases is on goal if we’re in Q3 of the year and there hasn’t been a release since last year. Do you consider the letter releases to count against the four quarterly source releases?
Eric,
Although you may think it’s not appropriate, we count letter releases (triggered by security releases) as releases — towards the at-least-twice-a-year binary target. That’s what I meant by “2. There will be at least two yearly ‘mature GA’ (currently MySQL 5.0) binary builds. They aren’t scheduled, but usually triggered by grave security vulnerabilities.” in http://blogs.mysql.com/kaj/2007/08/08/refining-mysql-community-server/
As for being on goal, by my comment “Also the four yearly source releases goal is on” in this blog posting, I didn’t mean we’re “on target”, I just meant “we haven’t skipped the goal”. Moreover, I confessed “We may need to improve on their predictability”.
Thanks for prompting this clarification!
Two things…
1. This is not how Open Source works. OSS projects don’t hide binaries from the community – at least not healthy ones.
Perhaps it’s time to stop pretending that MySQL is OSS. At best it’s source under glass.
(unless this policy radically changes while at Sun).
2. The community is less than happy here. Perhaps someone should host a public vote of no confidence in MySQL/Sun’s management of MySQL releases.
I don’t want to run stale versions of MySQL nor do I want to run versions that haven’t been tested by the community.
Damned if you do – damned if you don’t.
Kevin
Kevin,
Absolutely. I’ve made this argument on my personal blog too. And what I said before is still true: I WILL NOT PAY for what they are selling. I’d pay for something that’s good: but I won’t pay to expose my servers to risky, untested software. That’s just stupid.
The thing I just don’t get is that MySQL is turning up its nose at an opportunity to make a lot of money while making people happy and delivering a better quality product. All the blogs, interviews etc say “Our customers won’t pay for something they can get for free. That gives us two options: charge for services, or reserve some things (hot fixes, some features^H^H^H^H^H^H^H^H^H) for paying customers. The charge-for-services model has been tried and doesn’t work.”
This is a false dichotomy that IMO is perpetuated and pushed by the sales and marketing organizations within the corporation. There are more than 2 ways to make money, and the incredibly obvious one the community and lots of pundits have politely suggested time after time has been ignored.
If anyone else wants to make a good business out of this, there’s a great opportunity to do so. Since the source is open, anyone with the skills could do sane branch management and release a tested, stable version of the product. It would be a lot of work. That’s the point. That’s why it’s worth paying for.
Kaj,
How can letter releases count against the four yearly releases. It says right on the download page for 5.0.51b, “This is a bugfix release for the current production release family. It replaces MySQL 5.0.51.” Also, don’t hide behind the veil of security. Only one of the last two letter releases was for security and it wasn’t even due to security in mysql but the bundled yassl library. The other was due to mac os preference pane issues introduced by leopard. In either release _nothing changed_ in the core of mysql. I understand you can change the text of a README file, tar it up, and call it a new release but you’re not being very honest with us in doing so.
Please call this what it is, sunsetting the community release. I understand you’re trying to get the community to start their own releases. I don’t understand why. You already make a release, why force us to fork the code? We’re perfectly willing to work to write patches and make your release better, for _free_. All you have to do is accept some of the patches, keep us informed of your developments, and make releases for consistency and testing.
Peter, I’m new to MySQL. I have some questions.
1. What are the differences between MySQL Enterprise Server downloaded from provenscaling.com and the one bought from mysql.com? Can we download MySQL Enterprise Server from MySQL.com (without buying their MySQL Enterprise product)?
2. I read from MySQL.com, “MySQL Enterprise includes: MySQL Enterprise Server, MySQL Enterprise Monitor, MySQL Production Support”. So the only software that not available unless we buy MySQL Enterprise is MySQL Enterprise Monitor. Is it true?
3. What do you use to substitue MySQL Enterprise Monitor? Do you use MONyog monitoring tool?
Bob,
1) These are the same binaries. ProvenScaling does not rebuild binaries themselves but simply republish enterprise binaries which they allowed to do because they are distributed under GPL license.
2) Yes the Enterprise Monitor is only SOFTWARE though Support is the main value you get with subscription (especially as you’re beginner). There may be other software released in the future for enterprise users only.
3) MONYog is commercial tool as well though it is priced more fairly in my opinion (though the problem is mainly you can’t get MySQL Enterprise Monitor just by itself from MySQL). We use set of open source tools – NAGIOS, CACTI, Maatkit instead.
What is the advantages of using MySQL Enterprise edition compared to MySQL Community edition? Where can I get _trusted_ comparison of them?
Peter> ProvenScaling does not rebuild binaries themselves but simply republish enterprise binaries which they allowed to do because they are distributed under GPL license.
So they don’t include their “patches” like the one made by Percona? http://www.mysqlperformanceblog.com/2008/07/16/mysql-releases-with-percona-patches/
What’s a “trusted” comparison? Trusted by whom?
If “trusted” means “it’s worth something because I’m paying for it,” you can pay us to give you our opinion. A lot of people trust that pretty highly. Of course, you can just read our blog and get it that way, too.
Otherwise, you’ll have to make your own comparison. There’s nothing anyone can just hand you that we’re going to publicly say on our blog is trustworthy.