Will Facebook, Twitter, LinkedIn Stay With MySQL?

littlekorea writes "The world's largest web-scale users of MySQL have committed to one further upgrade to the Oracle-controlled database — but Facebook and Twitter are also eyeing off more open options from MariaDB and cheaper options from the NoSQL community. Who will pay for MySQL enterprise licenses into the future?"

and so meanwhile...

[Anonymous Coward]

... PostgreSQL is over in the corner, saying, "Hey guys! I'm open! I'm open!"

But no one throws the ball the Postgres. Because no one like Postgres.

So Postgres goes home and does some homework.

Government

[armanox]

Government for one. The US Department of Energy still uses MySQL, and I doubt they'll move off it anytime soon.

Re:and so meanwhile...

[0123456]

Funny, but not actually true.

We used to use MySQL unless a customer demanded Oracle. Now we've switched to Postgres, because MySQL's future is so hazy and we typically have to support these systems for ten years or more.

Re: and so meanwhile...

[Simon Brooke]

The real joke of this is that Postgres has been, by any measure, a better database than MySQL for twenty years. Back in the early 1990s when we were running on i386s and Sparcs, there was some argument for using MySQL because (in those days) the fact that it didn't have proper transactions and proper reverential integrity, it was faster for simple queries from single tables. Now, even that isn't true any more. Postgres is just the best engineered RDBMS out there bar none, and it's free.

Re: and so meanwhile...

[Anonymous Coward]

No, the real joke is Firebird DB is better than both MySQL and Postgres in terms of speed and disk usage.

Nobody uses it though. Firebird is to MySQL as BSD is to Linux. That is, it had some commercial/legal "complications" in the beginning of its life that have forever made it a loser despite being better.

Re:and so meanwhile...

[Maow]

I think what's missing is an easy upgrade path from MySQL to PostgreSQL.

For example:

• * mysqldump | psql doesn't work even with --compatibe=postgresql: ints have precision (int(11)) and comments don't work the same
• * Inside psql there isn't a handy "show create table" feature (that I've found)
• * No way to "use dbname" for switching DBs inside psql - must quit and restart with different dbname
• * Issues with double quotes vs single quotes vs ticks - no opinion on which is best way to go but would be nice if a translation were available
• * The commands aren't as easily memorable: \d vs show tables: another area where some compatibility would be nice. I kind of prefer the show tables, show databases, show create table style instead of \d, \l, \(can't do it in psql, use pg_dump)

Those are some things off of the top of my head.

Makes it so much more work to switch - each dumped table must be manually tweaked to load into psql.

I'm playing with it now, and growing more comfortable with psql but not sure I'm going to dump, edit, import all tables in all|any databases so I can have... 2 db servers running on my box.

I'm itching for a good reason to switch.

It's a shame that the new recently that Google is dropping MySQL didn't end with "and they're going to use Postgres" -- they have the resources to make a conversion suite / patches that would make it easy for a large scale adoption to occur.

Re:Licenses are a scam

[gagol]

It should be like cars, you can rent and pay base fee plus usage, or buy it and its yours.

Postgresql real problem

[petermp]

I personally think that the real problem with Postgresql happened 10 years ago. At that time it was not possible to run Postgresql on Windows(it was only possible via cygwin). That helped mysql get critical mass and Postgresql stayed behind. Then the snowball effect came into play and mysql was getting much more users compared to Postgresql.

Re:and so meanwhile...

[greg1104]

I made no such quoting error--that was direct from the New BSD text on the Wikipedia page--and I can't make any sense of whatever it is you're claiming. The 3rd clause of the New BSD license is "Neither the name of the nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission". That puts no restrictions on the code, only on the identity of the contributor. It just says that if I contribute code, the project can't use my name and write "Greg says our project is awesome" simply because I contributed--not without my written permission. That's what endorsement/promotion means here. The BSD licenses have always had "endorse or promote" restriction text of this form in them, because the University of California @ Berkeley didn't want people to think a BSD license says they approve of a program.

If you're reading any sort of profit or sale restriction out of that clause, you're very confused about what these licenses mean. I'd recommend Why you should use a BSD style license [freebsd.org] as a good piece comparing these licenses. Modified BSD License [oss-watch.ac.uk] is more terse description of the same area, with a particularly easy to follow description of New BSD->GPL moves work.

MariaDB picked New BSD as the alternate license because it's "GPL compatible". That means they can just slurp up any contributions under those terms without worrying about the copyright trail on that code at all. All they have to do is include the New BSD license in their source and binary distributions for those parts, not mention their contributors by name so they're not seen as endorsing that commercial version, and they're done. New BSD code gets assimilated trivially, GPL code comes in with copyright assignment, and therefore at all times MariaDB is uniquely able to sell derived products or the company itself with full ownership of any new code added--again.