Slashdot Mirror
Wikipedia Moved To MariaDB 5.5
Peetke writes "As we all know Oracle is not the biggest friend to the Open Source Community. Long standing OSS supporter Wikipedia has now moved from an optimized fork of MySQL 5.1 to MariaDB 5.5, for both its English and German sites. Wikipedia expects all other languages to follow within a month. Performance-wise, this move has no big implications, but it will ensure our biggest community database will live long and prosper."
For more information, Wikipedia has a statement regarding MariaDB: http://en.wikipedia.org/wiki/MariaDB
+1 IDisagreeSoHeMustBeATrollOrAnAstroturferOrAShill
Is slashdot next?!?!?!
RTFA.
It's part performance and part philosophical. Given that wikipedia is a strongly philosophical enterprise, this seems reasonable.
Replacing one database named after one of the author's daughters, with a database named after another of the author's daughters. Seems pretty consistent to me.
"As we all know Oracle is not the biggest friend to the Open Source Community." This is a bit weasely. We all don't know any such thing. For example, Oracle was in the top ten of organisations that contributed code to Linux last year: http://www.linuxfoundation.org/news-media/announcements/2012/04/linux-foundation-releases-annual-linux-development-report Since then it has been very public with Oracle Linux, and made several large contributions from that front. Shucks, it's even got its own OSS portal: https://oss.oracle.com/ I'm happy to agree it's a big bad corporate beast and does a lot of wrong in the world, but if you're going to criticize it, at least be factual.
"Oracle may screw MySQL".
Is there a reason for this other than ifs, buts and maybes?
One definition of madness is to try the same thing again and again and keep expecting different results. It's Oracle. You will get screwed.
SJW n. One who posts facts.
It's part performance and part philosophical. Given that wikipedia is a strongly philosophical enterprise, this seems reasonable.
Well, the performance difference didn't seem to be huge - in fact, some stats were slower.... I don't buy for a second that it was for performance reasons.
Philosophy - maybe - however Oracle contribute quite a bit to OSS - more than a lot of companies - See: http://www.oracle.com/technetwork/server-storage/linux/technical-contributions-1689636.html
In a nutshell, they are working on NFS over IPv6, data integrity checks for ext3, they maintain libstdc++, they worked hard on BTRFS, If anything, they have helped open source much more than most other companies.
Again, I don't see the philosophical reasons other than 'because we can'.
Sendmail is like emacs: A nice operating system, but missing an editor and a MTA.
Java was already open source when Oracle bought Sun. And since then, Oracle has been trying to close it back again with bullshit patent claims.
The first daughter's name is MySQL? I should introduce her to my nephew, little Tommy ;'Drop Tables.
- For the complete works of Shakespeare: cat
Her name is just My.
Right, they couldn't possibly have a good reason for doing it.
So an organization who asks for donations, waste their money changing Database systems for the sole purpose that they didn't like the company that bought the old one, although they didn't show any signs that they are going to damage the product or make it worse for them in any ways? Sounds like a wast of donated money to me.
So you didn't RTFA???
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.
Better performance on such a heavy traffic site is neither a waste of time nor money! ;-)
Since Oracle took over MySQL they've shown they routinely delay releasing patches for CVE security flaws for months until they can all be released together without documenting what fixes what issue. Several times updates have either ignore issues, removed fixes to earlier ones or in at least on case I remember applied a fix for an issue which failed to fix it and actually introduced a new one. This despite multiple FOSS projects (Debian,RHEL,MariaDB,Percona) having developed working patches separately which Oracle chose not to use.
They also don't disclose the details of many security vulnerabilities. That sounds reasonable on first glance, but it makes it a nightmare for sysadmins to assess whether it is worth system downtime to apply a patch, especially when that means upgrading to a newer DB version not tested against the application and which might break the application in several cases (for example due to the newer reserved keywords lists). A firewall or other measures may be sufficient to mitigate the threat, but that can't be assessed without seeing the details.
Ummmm... that's not what happened. They weren't using a stock release of MySQL. They were using an old 5.1 fork that Facebook created and has been maintaining. They decided they wanted the enhancements that the newer releases offered, and had a choice of migrating to a newer release of MySQL, or migrating to a newer release of MariaDB. Either way, they were migrating and had to put forth the effort to do so.
"Always forgive your enemies; nothing annoys them so much." - Oscar Wilde
Wikipedia was using a non-Oracle fork of MySQL (a Facebook maintained fork of MySQL 5.1) and moved to a different non-Oracle fork (MariaDB). The comment about Oracle not being a friend of OSS seems to be a non-sequitur.
Even if they are volunteers they could be working on doing something else that may have saved them money.
Have you ever managed volunteers? They work on what they want to work on. You can't just reassign them to a task they don't like, or they will walk away and donate their time to some other organization. Most likely they had some MariaDB fanboi (or fangoil) who was willing to do this, but was not willing to upgrade MySQL instead.
He also has a son named Max: http://en.wikipedia.org/wiki/MaxDB
Not only this (but please mod up anyway!), but as far as I know MariaDB is compatible with plugins designed for a comparable version of MySQL. At least for my Django and PHP work this holds true. I mean, isn't this the reason most developers abstract the database library anyway?
"Never let your sense of morals prevent you from doing what is right" - Salvor Hardin
If well MariaDB is backward compatible with MySQL, have some advantages on its own, like more choices for storage engines (i.e. Aria as a better myisam than myisam, xtradb instead of innodb, and others), and should have better performance in general than Mysql for the same equivalent version in the same hardware.
That Oracle is being bad right now with their concept of "open" (like suing Google for using Java) is an extra motivation.
Doubtful. More likely they wanted to be able to get decent community support for the forseeable future. Something that's not a given for a previously community-based software product that got gobbled up by a succession of commercial entities.
Oracle has gone to great lengths to make MySQL a second-class citizen to its own database in terms of support, and worse, they're not really getting the whole community part of why people used MySQL in the first place...or maybe they *do* get it and just want MySQL to go away so they can sell Oracle DB.
Either way, the folks at Wikipedia must have seen value in moving to a compatible, open-source, community-based database...just like the one they started with.
"Murphy was an optimist" - O'Toole's commentary on Murphy's Law