Wikipedia Adopts MariaDB | Hacker News Reader

Wikipedia Adopts MariaDB

blog.wikimedia.org

151 pointsJeremyMorgan38 comments

Official announcement: https://blog.wikimedia.org/2013/04/22/wikipedia-adopts-maria...

MariaDB looks likely to inherit a big slice, possibly even a majority, of MySQL's user base over time.

It's not just because MariaDB is as good as any of the "official" editions of MySQL offered by Oracle, but because it has only a single code base under a single Free license (as opposed to differently licensed "open source" and "enterprise" editions), because it has an active not-for-profit foundation looking out for the project and the community[1], and because many of the original contributors to MySQL, including founder Monty Widenius, are now working full-time on MariaDB.[2]

Say hello to the new MySQL -- MariaDB -- same as the old MySQL!

[1] https://mariadb.org/en/foundation/

[2] http://www.skysql.com/news-and-events/press-releases/skysql-...

The official post is excellent, we should swap out the blogspam link for it.

"For our most common query type, 95th percentile times over an 8-hour period dropped from 56ms to 43ms and the average from 15.4ms to 12.7ms. 50th percentile times remained a bit better with the 5.1-facebook build over the sample period, 0.185ms vs. 0.194ms. Many query types were 4-15% faster with MariaDB 5.5.30 under production load, a few were 5% slower, and nothing appeared aberrant beyond those bounds."

Ever since Oracle has taken hold of MySQL (I was at Oracle at the time, and saw their goals firsthand), they have been trying to extract as much money as possible from anyone using the database in a commercial way. Just the mention of MySQL strikes fear in the heart of any executive who has had any interaction with the mob that is the Oracle money extraction machine.

The sooner we all can move on and keep from using MySQL, the better, and good riddance to it and the red horse it unfortunately rides in on.

This article has many facts wrong.

It seems that MySQL development halted entirely at Sun.

I was concerned about Oracle's motives w.r.t MySQL but I can say they have come out with major new releases that improve MySQL in substantial ways. For instance, they've greatly improved the scalability of InnoDB on large SMP machines.

On the other hand, neither Sun nor Oracle have been really wanting to win with MySQL either and I think we could have seen more and different innovation in MySQL if it had been in other hands.

The Wikipedia win is big for MariaDB because it's a show of confidence that it works in demanding applications.

Was there ever a doubt that MariaDB wouldn't work in demanding applications? It's pretty much MySQL all the way down.

"No one ever got fired for buying IBM."

As a corollary, no one wants to be the guy that suggested MongoDB before it caused an outage.

I think that only makes sense if you don't know what MariaDB is. It's a fork that attempts to maintain high compatibility, run by the founder of MySQL.

What it isn't is a different program entirely, with a different protocol, syntax, scalability, consistency and community.

You missed the point. The idea is management and investor confidence, buy-in.

Here's how it works in most companies: MySQL goes down = something seriously bad happened. MariaDB went down = that database that Ted suggested sure is a piece of shit.

Well that depends on how Ted presented it, doesn't it? If he says "Let's try this other database instead of MySQL" may turn out different than if he says "Let's try this other version of MySQL put out by MariaDB".

In the end it comes down to having data to support your position though. If MariaDB goes down, and you can't answer whether it would have happened the same in MySQL, then maybe you shouldn't be making database suggestions. If it wouldn't have happened to MySQL, well then people were right to be wary, weren't they?

There are situations where MongoDB and the like are specifically more appropriate than traditional databases, and I do not include not understanding SQL syntax. In such situations where a ratio of needed scale to the resources available is particularly high, MySQL and other traditional databases are not a real option anyway.

However, there are of course also situations where, if you try to use MongoDB as a drop in replacement for a traditional database and then are surprised about data loss in exotic situations, for example, then you weren't really doing your job in the first place if the different goals never came up in your research of solutions.

It doesn't work for my employer's application.

May be the lots and lots of SPs are too much for MariaDB, but no, it is not a 100% replacement drop in.

I have tried.

Is there a reason to choose MariaDB over Percona or vice-versa? They're both drop-in replacements for MySQL and both replace InnoDB with XtraDB (which Percona makes).

The only real feature MariaDB offers over vanilla MySQL and Percona is an improved optimizer for complex sql queries [1].

In most cases Percona seems to be a win over both MySQL and MariaDB as it is just patch layer over MySQL with features and tweaks that real production users see and need. These features tend to end up in future MySQL releases like saving and restoring the buffer pool allowing mysql to warm start. I believe that feature made it in to MySQL 5.6 but was added to percona/mysql 5.1.

[1] http://blog.mariadb.org/mariadb-5-3-optimizer-benchmark/

Is MariaDB still substantially slower than MySQL?

http://dimitrik.free.fr/blog/archives/2013/02/mysql-performa...

The official blog post makes it sound like overall it was a performance gain with a few edge cases coming out 5% slower.

I wish it talked more about their DB hardware, I would believe MariaDB is more performant for scale out, whereas the MySQL improvement have been targeting scale up (something Oracle is very good at).

The big battle MariaDB (or any DB) is going to have is gaining ground in the shared web host space. That makes up a pretty large portion of websites, and there is no incentive for those providers to change. I'd love to see other DB options in that space, its all PHP/MySQL.

That'll probably happen - mainly because the distributions will probably switch over to MariaDB. I'd expect Ubuntu, Debian, CentOS etc to all ship MariaDB as an upgrade at some point.

Unfortunately it would appear Debian is actually lagging a bit behind in this regard:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=565308

The only problem I foresee with MariaDB is they have no interest in porting innodb fulltext search like MySQL 5.6

Sure there are other fulltext solutions but to simply convert myisam to innodb, FTS on innodb with the same queries is rather helpful.

So bets on how long it will be until MariaDB is scooped by Oracle?

I had the benefit of seeing Monty talk at All Your Base conference in Oxford last year, and let me tell you, he made it quite clear that a huge driving force behind MariaDB is to take on Oracle, because he absolutely hates them for what they've done with MySQL. Remember, the original sale of MySQL wasn't to Oracle, it was to Sun, which was then acquired by Oracle. After hearing what he had to say, I have nothing but a tremendous amount of respect for Monty, and wish him all the success in the world.

Except, as much I am not a fan of the company, Oracle has done a far better job managing and improving MySQL than Sun so far.

Having the same company own both the main engine (InnoDB) and the harness (MySQL) has been tremendously valuable.

> he absolutely hates them for what they've done with MySQL

I may be a bit ignorant, but what have they done with MySQL?

You can watch the video of his talk here. http://vimeo.com/56639635

Perhaps "hate" is the wrong word, and it should be more "dislike", since Oracle are essentially the complete opposite of what Monty wants to encourage (e.g. Hacker Business Model, Open Source). My memory's a bit hazy, but from what I gathered, Oracle tried to stop MySQL in its early days by acquiring the INNODB Engine around 2005. After that, once Oracle acquired Sun and thus MySQL, Monty and many of the other original developers saw Oracle as being against the spirit of the original intention of MySQL, shifting away from the principles they wanted to encourage, particularly open source philosophies, and they wanted to make sure a version of MySQL would always stay free. From what I recall, he believed that Oracle were no longer fixing known bugs because they didn't know how, and they were also moving from "open source" to "open core", where new features were closed source.

I really hate the word hate. This has the whole stupid jenkins/hudson thing written all over it.

I do appreciate Monty's efforts, but really, can we kiss and make up? And if oracle wants to monetize MySql, so what? The rest of us get a better database.

Previous release of MySQL new version was really evolutionary, so I'm sure now that Oracle can improve this DB. And if we will add some facts about InnoDB then all Monty's "hate" will look like populism or try to keep the face.

...after he made wheelbarrows full of money from the Sun purchase.

MariaDB basically exists because of Monty's hatred for Oracle (which bought Sun after Sun bought MySQL for wheelbarrows full of money from Monty). It won't happen.

If it was me, I'd keep cranking out superior databases, selling them for a billion, and starting on the next one as fast as they'd buy em :)

I think it would be a bit premature to call MariaDB superior to MySQL.

Also it should be noted that the InnoDB storage engine was developed outside of MySQL AB. InnoDB is a helluva lot better than MyISAM for most applications because of performance and support for transactions, consistency, etc. Without it I don't think that MySQL would be a good candidate for many applications, particularly given that Postgres has improved a lot. I.e. I'm not sure Monty should really get credit for everything about MySQL/MariaDB.

I think that's what Stonebreaker did with PostgreSQL, VoltDB, Vertica...

Not sure if intentional, but Stonebreaker is an awesome misspelling of Stonebraker.

Yes, I misspelled Stonebraker, thanks for the correction.

Monty started MariaDB pretty much as soon as he left Sun, long before the Oracle acquisition, but shortly after he started criticizing the release of mysql 5.1 under Sun, which was widely regarded as botched.

It would be an interesting move from Oracle to even try. Nothing would say corporate desperation as much as trying to buy the project that sprung out when you first bought the project.

That said, I wouldn't put it beneath them...

It does not make sense for Oracle to try that. MariaDB demonstrates there is enough support to sustain a DB that is MySQL but without Oracle. They'd have to be dumb not to expect another MariaDB to spring up after the hypothetical takeover.

Not that the takeover is feasible, as the other answers point out.

Anyone know of any hosted MariaDB offerings out there that can be used with Heroku?

Reply on news.ycombinator.com