Is MySQL becoming less open source? Test cases go internal now | Hacker News Reader

Is MySQL becoming less open source? Test cases go internal now

blog.mariadb.org

93 pointsbytebot42 comments

So basically, the Java strategy. Killing the alternatives through creating compatibility uncertainty. Oh, man. This doesn't bode well.

I hope this drives developers towards using and improving Postgres, but, well, who knows. One can dream!

More likely is that it may drive people towards Percona, which I have on good word is a solid professional MySQL branch...or, like Java, people will just stick with the brand. :/

EDIT: adding links for the curious.

Percona: http://www.percona.com/

Background on the Java stuff: http://arstechnica.com/information-technology/2010/12/apache... (the difference is that MySQL never did have open governance, so there was nobody to quit in a justified huff)

For the majority of developers, it's whatever database server that control panels like cPanel and Plesk choose, and they aren't the sort to switch on a whim.

For anyone interested: cPanel doesn't officially support PgSQL, and Plesk charges $6-15/mo for their "Power Pack" addon which adds PgSQL support.

I would think one would need to look no further than the Oracle/Google lawsuit and its outcome with respect to Java to see how Oracle will move forward with MySQL.

Java has been greatly improved since Oracle took over the helm. Major improvements to the JVM, an exciting yet stable roadmap, better platform support (in particular on OSX) and a lot more confidence about its future.

Oracle has been nothing but a huge positive for the Java platform.

> Oracle has been nothing but a huge positive for the Java platform. For the OpenJDK Java implementation, definitely. For free and open alternatives? No, not at all.

While it has been positive indeed, I wouldn't say it "has been nothing but a huge positive".

The closing down (of the JCP, the lawsuits, etc) in particular has been a little worrisome.

No one involved with mySQL needs pgsql to improve, they just need a couple of days to learn pg config. Pgsql is already a much more sophisticated and debugged database.

MySQL is as popular as it is because it has been free (as in beer), and relatively easy to set up and use.

The time for that has come and gone. MySQL's shortcomings limit its use in many corporate and professional settings, and the alternatives are now much easier to use.

If I were looking to become a DBA today, I would study PostgreSQL. Likewise, if I were developing a new web application, MySQL wouldn't even make my list.

MySQL is as popular as it is because EVERYONE is using it.

Which means that all of the really hard scalability and management problems have already been solved every which way. And all available for free by companies you know and trust e.g. Facebook, Twitter.

If I were looking to become a DBA today I wouldn't bother with PostgreSQL at all. I would learn all of the NoSQL databases because that is where all of the interest is right now. Due to the increase use of agile in software development databases need to be flexible with the schema controlled by developers not DBAs.

Or you could, you know, learn multiple things, because very few companies that need scalability are best served by doing everything in only one technology. Or, for that matter, hiring people who only know one technology.

There's a hell of a lot of useful stuff to learn (practical and theoretical) from relational databases. NoSQL has some useful ideas too. However, advocating that someone wanting to be considered a "database administrator" only learn/focus on NoSQL - that's just poor advice.

Funny that you chose to reply to me instead of the parent. Learning just PostgreSQL is also poor advice.

Most DBAs will already know Oracle and MySQL and my point was that they would be much better advised learning a NoSQL database. Of course if you're a new DBA you should learn a relational database I just fail to see why on earth anybody would recommend PostgreSQL over Oracle/MySQL which are far, far, more popular.

"If I were looking to become a DBA today I wouldn't bother with PostgreSQL at all. I would learn all of the NoSQL databases because that is where all of the interest is right now."

That's what I was responding to. "looking to become a DBA" does not remotely imply someone who already "knows" Oracle and MySQL.

Oracle might be popular, but few places can afford, or justify paying for it. MySQL is mostly popular because everyone was busy cargo culting based on what Facebook and YouTube were doing instead of evaluating the choices on their merits. Luckily most of those people have moved on to cargo culting NoSQL choices.

If you want a solid open-source database, especially one that has a future, learn Postgres.

I love armchair experts.

So I guess if all of those clueless engineers at the world's most trafficked websites would have just listened to you they would never have picked MySQL. Have you ever thought of letting them know that they should switch over to PostgreSQL ?

The engineers and Facebook might have selected MySQL on merits. Or they might be stuck with trying to do the best with legacy choices. I don't know what the best solution is for them today, and it might not be either Postgres or MySQL.

If however you need to make a choice today for what database to use on a new project, then you should evaluate the choices instead of basing your decision on what might have been best for Facebook 8 years ago.

I'm not going to get into the technical differences here, because even if MySQL and Postgres were technically equivalent (and they aren't but I'll let you do your own research), I simply don't trust Oracle with the future of MySQL.

>I love armchair experts.

Speaking as a battle-scarred, proven old sage veteran of the DB wars? Because the opinions you state above don't really qualify you as that...

To be clear, I did not advocate learning only anything.

If you aren't interested in learning about MongoDB, MySQL, DB2, and stack of brittle punch cards, you're probably not cut out to be an IT professional in the first place.

dba and nosql? sounds strange. note that scalability is not everything.

It's strange a DBA managing a database ?

And you're right. Scalability isn't everything hence the popularity of NoSQL which are far more developer friendly than current SQL databases. And in most companies it is the development teams that dictate technology choices.

This is a MySQL discussion, but I think it's worth it to say what two things I think keep Postgres from becoming the viable worldwide use-by-default contender that MySQL became:

1. Branding. First and foremost. MySQL had great documentation, an identifiable logo, and a wonderful name that every non-technical person could remember and identify. Postgresql is just, well, it's a horrible nightmare of a name. People would like to think this doesn't matter but: of course it does.

2. A really slick installer for Windows. You could say this is branding and to some extent it is, but MySQL and its admin tools have a really friendly installer that lets intermediate developers without a lot of DB experience get it up and running with their development environments quickly. I admit it's been a while since I used Postgres's but the fit and polish wasn't there.

this comment is not only misplaced, it is (at least partly) wrong as well. postgres has great documentation. and had great documentation for quiet a while. see [1].

[1] http://www.postgresql.org/docs/7.3/static/

Let's keep this on topic: in what specific ways has postgresql been more open, and how does that lead to a better developer community and greater value for users?

I happen to think the postgres docs are great, and they are open and free unlike the MySQL docs. I have tried following one MySQL doc, only to see it change later and have no way to see the old version. Not a problem for postgres.

If you have some suggestions for the docs please submit them to the docs mailing list.

Postgres also has 15+ years of history in git, which is extremely helpful.

Honestly, I don't know where to start. Postgres is an open community project to the core and benefits from that in so many ways.

http://www.engineyard.com/blog/2012/postgresql-is-our-new-de...

Despite the reasons they gave, did they see the writing on the wall that made this the time to make the leap?

I haven't yet used one of the recent MySQL forks, but I have a feeling the next time I'm using an OSS solution that requires a MySQL variant, I wont be using MySQL.

I was recently looking at the alternatives available, and thought that Drizzle [http://www.drizzle.org/] looks really great.

Drizzle is not a drop-in replacement for MySQL. The only real option if you want to replace MySQL without having to change your application code is MariaDB. Drizzle is fine for new projects, but you'd have to massively alter an existing site to get it to work with Drizzle.

I don't think it's fair to say MariaDB is the only real option. The CTO of the company I work at seems to be a pretty firm advocate for Percona.

I think MariaDB should be a viable alternative, right?

Absolutely. It's been a viable alternative for some time now.

So, time for LibreSQL?

or YourSQL

MariaDB is the free (GPL) branch (actually more of a fork at this point) created by Monty, the primary original author of MySQL.

Is anyone actually using MariaDB in production at a reasonable scale? Our DBAs tried to convince us to switch from MySQL to Maria a few months ago. All of the devs on my team looked around and couldn't find any testimonials, use-cases or benchmarks that weren't done by someone on the Maria dev team. I realize Monty is behind it but I don't really feel like being an early adopter based on fear that some day Oracle might do something bad with MySQL.

I use MariaDB on all of my websites these days, the largest of which tops out at around 5K queries a second. The biggest draw for me was the XtraDB engine (which is also found in Percona Server) which is a replacement for InnoDB. XtraDB performs much better on SSDs and I was able to get over 50% performance gains on some queries with the default settings. This may or may not be applicable to your specific environment, but it is a drop-in replacement so it's very easy to just install it on a test box and stress test it with your current code.

To be clear, XtraDB is solely developed by Percona. MariaDB has nicely included it, not the other way around.

I was just pointing out that Percona Server is another MySQL drop-in replacement since the commenter seems to be asking about an enterprise environment which Percona Server is targeted more towards than MariaDB. You're right I should have clarified that XtraDB is a product of Percona.

I've been using it in production as the primary datastore on a news website with fairly high traffic (10 million+ monthly uniques, 1000s of queries per second). Have not encountered any issues since switching to MariaDB over a year ago. Monty's organization places a high value on well tested and stable releases.

http://drupal.org/node/861192 comment #4 - used for Drupal's test cluster

http://kb.askmonty.org/en/mariadb-case-studies/ (including Limelight networks)

Switching in early 2010 would have made you an early adopter.

My sense is that most sites use Percona Server instead of MariaDB, because Percona and xtradb (mods to innodb) have been around longer, and those changes are generally a subset of the changes MariaDB incorporates.

Not an expert here: Will all those PHP projects in the wild work with MariaDB? Everytime a friend needs a blog or wiki I have to make sure MySQL is installed because it seems few people who make PHP projects allow other backends.

I'd love to apt-get purge mysql-server from my servers.

Does someone know of a really good (performance + corner cases) migration tool for MySql - > Postgres.

It looks like there are a ton of roll-your-own, but none as good as, say, "git svn clone"

I can't really speak to performance or corner cases outside of our (admittedly small) needs, but one of our developers had good luck with mysql2postgres -- https://github.com/maxlapshin/mysql2postgres/

Reply on news.ycombinator.com