Will Facebook, Twitter, LinkedIn Stay With MySQL?

← Back to Stories (view on slashdot.org)

Will Facebook, Twitter, LinkedIn Stay With MySQL?

Posted by timothy on Sunday September 22, 2013 @09:17AM from the so-long-as-it's-internet-scale dept.

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?"

47 of 245 comments (clear)

Min score:

Reason:

Sort:

and so meanwhile... by Anonymous Coward · 2013-09-22 09:20 · Score: 5, Interesting

... 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.
1. Re:and so meanwhile... by 0123456 · 2013-09-22 09:40 · Score: 4, Interesting
  
  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.
2. Re:and so meanwhile... by Anonymous Coward · 2013-09-22 09:46 · Score: 5, Insightful
  
  PostgreSQL's biggest disadvantage over MariaDB is that it's not a drop-in replacement for MySQL.
3. Re: and so meanwhile... by Simon+Brooke · 2013-09-22 09:46 · Score: 5, Interesting
  
  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.
  
  --
  I'm old enough to remember when discussions on Slashdot were well informed.
4. Re: and so meanwhile... by buchner.johannes · 2013-09-22 09:55 · Score: 5, Funny
  
  Maybe Postgres lacks discoverability?
  
  --
  NB: The message above might reflect my opinion right now, but not necessarily tomorrow or next year.
5. Re:and so meanwhile... by pspahn · 2013-09-22 10:32 · Score: 3, Insightful
  
  That sounds like more of a disadvantage of the application itself rather than the DB platform.
  This is 2013... use an ORM.
  
  --
  Someone flopped a steamer in the gene pool.
6. Re: and so meanwhile... by Anonymous Coward · 2013-09-22 10:33 · Score: 5, Insightful
  
  How did MySQL get such critical mass?
  Probably the main reason is that it has a "design philosophy" of "if you can't do what the user wants, better to do something and say it's all OK than to give an error", which some people mistake for ease of use.
7. Re: and so meanwhile... by emilper · 2013-09-22 10:43 · Score: 2
  
  back in the early 90 MySQL and Postgresql did not exist
8. Re: and so meanwhile... by SQLGuru · 2013-09-22 10:43 · Score: 5, Funny
  
  BetaMax vs VHS. Blu-ray vs HD-DVD.
  Obviously it's because the porn industry chose MySQL.
9. Re: and so meanwhile... by Anonymous Coward · 2013-09-22 10:49 · Score: 2, Interesting
  
  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.
10. Re:and so meanwhile... by Maow · 2013-09-22 11:16 · Score: 3, Interesting
  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.
11. Re: and so meanwhile... by gman003 · 2013-09-22 11:17 · Score: 5, Informative
  
  Uh, Postgres has all the standard GRANT and REVOKE, plus some things I don't immediately recall MySQL supporting. This support goes back at least to 7.3, which is a decade old. From what I can tell from the changelogs, looks like they started adding that around 1997 in 6.0.
  I'll also note that PostgreSQL places a lot of importance on following the standards - they seem to support far more things than MySQL. In fact, looking at their "list of unsupported SQL features", it seems the bulk of them are "embedded [outdated programming language]" of one sort or another, or fancy XML stuff.
12. Re: and so meanwhile... by hibiki_r · 2013-09-22 11:29 · Score: 5, Insightful
  
  MySQL got its critical mass by it's easy, tight integration built into PHP. Any random college student could build a website backed by a database pretty quickly. It was a total failure to anyone that wanted to do serious work with it, but serious work was never an issue. As those college students entered the workforce, they tried to keep the tools they learned. People worked around their tech's limitations until new versions added it in, instead of migrating to competitors.
  So it was a perfect storm or filling a niche for a community that just kept growing.
13. Re:and so meanwhile... by Jane+Q.+Public · 2013-09-22 11:55 · Score: 3, Informative
  
  "Now we've switched to Postgres, because MySQL's future is so hazy..."
  It's no more hazy than it was when Oracle took it over. The MariaDB project is largely run by ex-MySQL developers... where's the problem? If anything, it was Oracle that muddied the waters. Now things are getting BACK on track.
  
  I like Postgres in some ways, but it has some significant deviations from standard SQL syntax, and other idiosyncracies.
  
  For me (I'm not doing anything "enterprise" at the moment), the slight performance gain of Postgres is not worth putting up with its oddities.
14. Re:and so meanwhile... by greg1104 · 2013-09-22 12:30 · Score: 5, Insightful
  
  As long as MariaDB is requiring copyright assignment, there's every reason to believe it will be sold off again the same way MySQL was. The FSF gets away with that for GNU projects because they've never abused contributor trust before. Monty is no FSF, and there's no reason believe MariaDB will remain outside of commercial control any better than MySQL did. I can't believe people are falling for the same trick again.
  PostgreSQL aims for SQL standards conformance as much as possible. It's hard sometimes due to the difficulty of participating in the standard process. The idea that MySQL does a better job in that area is kind of odd though. You'll have to list some sample Postgres "oddities" to be credible with that claim.
15. Re:and so meanwhile... by greg1104 · 2013-09-22 12:34 · Score: 5, Funny
  
  PostgreSQL's biggest disadvantage over MariaDB is that it's not a drop-in replacement for MySQL.
  Postgres has a blackhole engine that loses data in unpredictable ways now. But that only emulates parts of MyISAM.
16. Re:and so meanwhile... by hairyfeet · 2013-09-22 13:08 · Score: 4, Insightful
  
  What I don't get is this...why does anybody think MariaDB is ANY safer? I mean its still run by old Monty, right? The guy that sold MySQL out from under the community to Oracle in the first place? And don't he still require that ALL code contributed have the rights signed over to him?
  If he fools you once? Shame on him. If he fools you twice? You are an idiot that deserve what you get.
  
  --
  ACs don't waste your time replying, your posts are never seen by me.
17. Re: and so meanwhile... by turbidostato · 2013-09-22 18:36 · Score: 2
  
  "The network connections one has some merit."
  "Doesn't support" and "it doesn't configure itself by default to be cracked ala red worm" are quite different things. By the way, all sane installers of MySQL do exactly the same thing.
18. Re:and so meanwhile... by Chuck+Chunder · 2013-09-22 18:39 · Score: 2
  
  No way to "use dbname" for switching DBs inside psql - must quit and restart with different dbname
  /c dbname
  
  --
  Boffoonery - downloadable Comedy Benefit for Bletchley Park
19. Re:and so meanwhile... by Chuck+Chunder · 2013-09-22 18:40 · Score: 3, Insightful
  
  \c obviously
  
  --
  Boffoonery - downloadable Comedy Benefit for Bletchley Park
20. Re:and so meanwhile... by ianare · 2013-09-22 20:02 · Score: 3, Informative
  
  I like Postgres in some ways, but it has some significant deviations from standard SQL syntax, and other idiosyncracies.
  Strange you would mention that, one of the reasons I've switched to PostgreSQL (and never looked back) is because it more closesly follows the SQL standard and has many less "gotchas" and bugs than MySQL (boolean is actually an int field, reset counter on increment, etc).
  When people complain about Postgres' "non-standard SQL", this usually comes from those that have only used MySQL and think it's the standard.
  About the only technical advantage MySQL has over Postgres is an easier setup, and generally better performance out of the box (before any tuning).
21. Re:and so meanwhile... by Chrisq · 2013-09-22 21:12 · Score: 2
  
  The "problem" with Postgresql is the Postgresql developers/team are more likely to present a realistic picture of things.
  Those who really know about clustering will know it's all a bunch of compromises.
  Just because MySQL fans paint a rosier picture doesn't mean the MySQL situation is really that much better: http://bugs.mysql.com/bug.php?id=25543 https://dev.mysql.com/doc/refman/5.6/en/mysql-cluster-replication-issues.html
  MySQL has had replication as a built-in for a long time, and I encountered such a system at work when it was 3.x. BUT the clustering system seemed to cause more problems than it solved - stuff kept getting out of sync or the syncing completely stops. Was someone else's job to maintain it, but I had to help fix it from time to time...
  Might be better now, but point is MySQL implied it was fine back in 3.x despite its crappiness.
  I don't doubt what you say (I don't have the expertise to judge it) but I should have pointed out that the number of people who really need clustering (as opposed to hot or warm backup) is very small. We realised that beefier hardware with hot backup was a much easier solution. Our DB expert tells me that the number of people who need a scalable conventional SQL cluster is getting to be a really small niche, with single-node systems improving and eating into the bottom half and NOSQL systems eating into the top.
22. Re: and so meanwhile... by inglorion_on_the_net · 2013-09-22 22:37 · Score: 3, Informative
  
  Go back further to when MySQL got momentum and Postgres did not do SQL *AT ALL*.
  [citation needed]
  Actually, let me get some citations for you, although they contradict your statement:
  From: https://en.wikipedia.org/wiki/Postgresql
  
  In 1994, Berkeley graduate students Andrew Yu and Jolly Chen replaced the Ingres-based QUEL query language interpreter with one for the SQL query language
  From: https://en.wikipedia.org/wiki/Mysql
  
  The first version of MySQL appeared on 23 May 1995
  So it would appear that Postgres supported SQL before MySQL even existed.
  
  --
  Please correct me if I got my facts wrong.
23. Re:and so meanwhile... by Burning1 · 2013-09-22 23:11 · Score: 2
  
  The first lesson I learned from the book 'High Performance MySQL' is that one should never use an ORM. Optimizing for ORM queries is virtually impossible.
  http://stackoverflow.com/questions/7707976/php-and-mysql-high-traffic-solution
  FWIW: I notice in a later post you mention PDOs... Which look like they won't impact optimization strategies.
24. Re:and so meanwhile... by jonbryce · 2013-09-23 00:06 · Score: 2
  
  That can only happen if they replace all the MySQL code with something completely different. MariaDB doesn't own the commercial rights to that because they sold it to a company now owned by Oracle, only the same GPL rights that everyone else has.
25. Re:and so meanwhile... by lgw · 2013-09-23 05:23 · Score: 2
  
  If you're starting a large-scale project from scratch, by all means design it for NoSQL (and find some way to deal with the transactions you still need - you'll probably still end up with a few normal DBs for lock tracking, as distributed locking is hard). That's fairly useless for existing large scale apps: basically everyone but Google is using sharding over SQL DBs, and you just can't "port" between those worlds.
  Eventually I expect all the big players to do from-scratch re-writes for the NoSQL world, but not as a response to MySQL swirling the drain. That's something you do once you're product is 10+ years old, but you still see enough future work that it's worth a re-write. No one's quite there yet with cloud stuff, and the first will be Google (already NoSQL) and Microsoft (which will be interesting to watch).
  
  --
  Socialism: a lie told by totalitarians and believed by fools.
26. Re:and so meanwhile... by greg1104 · 2013-09-23 06:37 · Score: 2
  
  There were some contributors to MySQL AB code that were hired as employees. Other than that, community contributors got nothing from the eventual sale. MariaDB is structured the same way. The only way they share profits is if they're impressed enough to hire you. Otherwise community contributors got zero before, and so far all the evidence says they'll get zero if Monty sells out again.
27. Re:and so meanwhile... by greg1104 · 2013-09-23 13:32 · Score: 3, Interesting
  
  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 as a good piece comparing these licenses. Modified BSD License 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.
Government by armanox · 2013-09-22 09:24 · Score: 3, Interesting

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

--
I'm starting to think GNU is the problem with "GNU/Linux" these days.
1. Re:Government by confused+one · 2013-09-22 09:52 · Score: 2
  
  Are they paying Oracle for support? Or are they just using MySQL in their server farm, supporting it with internal resources?
Re:Enterprise? by Florian+Weimer · 2013-09-22 09:44 · Score: 5, Informative

And the article confirms the large-scaler users aren't part of that elusive group, either:

Many of the largest MySQL users — Twitter included — do not currently pay Oracle for an enterprise licence. Twitter, like Facebook, prefers to build their own extensions and customisations off the community version.
How do you get cheaper than free? by msobkow · 2013-09-22 09:52 · Score: 2

You don't have to pay for a commercial license of MySQL as far as I know, unless you want support for it.
And even if there were a dollar difference, I doubt it would be enough to cover the cost of redeveloping everything to use NoSQL servers.
Hell, it's not even cost effective to switch to another SQL database like PostgreSQL.
Can you imagine the downtime required to export Facebook from MySQL and to re-import it to another database? The users would go ballistic!
I don't expect any "earth shattering" movement by any of the big users in the near future.

--
I do not fail; I succeed at finding out what does not work.
1. Re:How do you get cheaper than free? by sgbett · 2013-09-22 10:06 · Score: 3, Funny
  
  "One does not simply export Facebook!"
  
  --
  Invaders must die
2. Re:How do you get cheaper than free? by rml1997 · 2013-09-22 10:13 · Score: 5, Informative
  
  Hell, it's not even cost effective to switch to another SQL database like PostgreSQL.
  Can you imagine the downtime required to export Facebook from MySQL and to re-import it to another database? The users would go ballistic!
  I don't expect any "earth shattering" movement by any of the big users in the near future.
  I'm involved in a project that involves moving databases. We write each transaction to both the old and new structure using our data access layer, then export historic data and eventually, once we've verified the new system is working as expected, remove the old structure from the data access layer. This is the main reason data access layers are used.
'looking at' NoSQL? by phantomfive · 2013-09-22 10:01 · Score: 2

That's strange, for some reason I had the idea that Twitter and Facebook were already using NoSQL. If they aren't, then is any large company using NoSQL?

--
"First they came for the slanderers and i said nothing."
1. Re: 'looking at' NoSQL? by jbo5112 · 2013-09-22 13:27 · Score: 3, Funny
  
  Facebook uses a NoSQL database (HBase) for their messaging system, and some related tech for data analysis (Hadoop). They also have custom photo serving software (haystack) for their photo storage. The main data (status updates, friends, likes, etc.) is in MySQL with memcache in front of it. There is also a cache layer (varnish) in front of the web servers. They said NoSQL isn't ready and point to the smaller messaging system needing more staff.
Re:"Will businesses needlessly give away money?" by NoNonAlphaCharsHere · 2013-09-22 10:05 · Score: 4, Insightful

Because the managers who select databases are buying ass-coverage. Performance and cost are secondary considerations.
Reverential integrity? by Anonymous Coward · 2013-09-22 10:20 · Score: 4, Funny

One entire Billy Graham at a time?
what about slashdot? by larry+bagina · 2013-09-22 10:27 · Score: 3, Funny

are you guys still using MySQL?

--
Do you even lift?
These aren't the 'roids you're looking for.
Re:"Web-scale"? by gl4ss · 2013-09-22 10:47 · Score: 2

as far as I gather it's a joke.
youtube for a video.
it has to be a joke. just has to!
(I guess it's supposed to mean that your site wont break even if you get slashdotted)

--
world was created 5 seconds before this post as it is.
Aquire and Obscure by Joebert · 2013-09-22 10:53 · Score: 2

I don't think Oracle really ever planned on doing much more with MySQL than keeping control of it until it dies.

--
Wanna fight ? Bend over, stick your head up your ass, and fight for air.
They should switch to Mongo DB by citizenr · 2013-09-22 11:21 · Score: 3, Funny

Its web scale!

--
Who logs in to gdm? Not I, said the duck.
Re:License by Anonymous Coward · 2013-09-22 11:44 · Score: 2, Informative

Bullshit.
It's a variation of the BSD/MIT license, which has fewer restrictions than the viral GPL. It's functionally equivalent to the most permissive Creative Commons license, only requiring attribution. It's explicitly listed on the OSI website as an approved license.
"Permission to use, copy, modify, and distribute this software and its documentation for any purpose, without fee, and without a written agreement is hereby granted, provided that the above copyright notice and this paragraph and the following two paragraphs appear in all copies."
The only people who object to this type of license are GPL bigots who want to impose their version of "freedom" on everyone else.
I switched to MariaDB by EmperorOfCanada · 2013-09-22 12:54 · Score: 3, Funny

I switched to MariaDB but my database is the size of a microbe so the few quirks were of no difficulty; but there were quirks. MariaDB was not a plug in replacement. I love it and wouldn't go back but it did take a tiny bit of work. So if I had one zillion servers with crazy databases I would be taking my time on that one. I suspect that what you will see is new development experiments depending on MariaDB and slowly increasing the pressure until they just make the switch.

The other question is how many obscure features of MySQL features are they using? (Including custom code)
Re:Licenses are a scam by gagol · 2013-09-22 12:57 · Score: 3, Interesting

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

--
Tomorrow is another day...
Postgresql real problem by petermp · 2013-09-22 17:59 · Score: 5, Interesting

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:Enterprise? by Bigbutt · 2013-09-23 00:44 · Score: 2

At the moment we do. We have a moderately sized Oracle environment but the company owners have been annoyed with the Oracle support costs and started moving to MySQL several years ago. Considering our environment, we paid for MySQL enterprise licenses. When Oracle bought MySQL, the company started moving to Postgres. Same when Oracle killed Sun (the Oracle DB license fees kept us from upgrading our older Sun equipment so we moved to Linux on Dell and then virtual machines). Now we're using Redhat and Postgres (it's all in the pipeline so there are still mySQL deployments).
[John]

--
Shit better not happen!