Feeds

back to article Red Hat to ditch MySQL for MariaDB in RHEL 7

In a surprise move, Red Hat has announced that version 7 of Red Hat Enterprise Linux (RHEL) will ship with the MariaDB database installed by default, in place of MySQL. The announcement was made at the company's Red Hat Summit, which wrapped up in Boston on Friday. MariaDB is a fork of MySQL that was launched in 2009 by …

COMMENTS

This topic is closed for new posts.

Page:

Silver badge
Pint

I dunno

Why not make a real statement and use (for example) PostGreSQL instead?

Each to his own, but it would have left a little more impression with me than simply going for the "best mysql compatible dba out there".

3
9
Anonymous Coward

Re: I dunno

And why would you not go for the "best mysql compatible dba out there".

5
0
Pint

Re: I dunno

RHEL does have PostgreSQL and has for some time; this change doesn't affect that.

10
0
Silver badge

Re: I dunno

That is rather like saying "we won't ship Perl because we ship Python instead. They are both scripting languages and you can do much the same thing in either." But that does not help if youhave an application written in Perl.

Similarly: there are many applications written to use Mysql so RedHat have to ship something that is Mysql compatible so that their uses can run them. Agreed: Postgresql might be better for a new project, but that is a different story.

9
1
Anonymous Coward

Re: I dunno

More 'Open Source' fragmentation that no one cares about. Usage of this product is near zero outside of web hosting and internet companies....

1
11
Silver badge

Re: Usage of this product is near zero outside of web hosting and internet companies

I think I remember people saying that about Linux too.

5
0
Anonymous Coward

Re: I dunno

PostGreSQL, hmm yes - why not make my servers even more owned than they usually would be running Open source....

Quite impressive to achieve a CVE level 10 vulnerability in a database: http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2013-1903

0
1
Anonymous Coward

Re: I dunno

Even more impressive to have 2 of them: http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2013-1902

0
1
Anonymous Coward

Re: I dunno

Not so impressive to have your DB server directly accessible from the Internet. Amateurs.

0
0
Silver badge

PostgreSQL

Agreed with ShelLuser: they should use PostgreSQL as the main DBMS. Monty is part of why I disliked MySQL and eventually turned back to PostgreSQL ... he hated transactions and thus made MyISAM and the early MySQL releases transactionless. Also, referential integrity was missing because of that.

And well, MariaDB is headed by Monty. Using the transactional stuff requires InnoDB which is owned by ... Oracle. Oops!

2
10
Silver badge
Linux

Re: PostgreSQL

Good decision, InnoDB went out the window years ago.

0
0
Anonymous Coward

Re: PostgreSQL

InnoDB has been GPL for some time as well as the default storage engine for some time, and MariaDB has already replaced MySQL on many other (uptodate) Linux distros

3
0
Anonymous Coward

Re: InnoDB

and of course XtraDB is also provided in MariaDB as a fork of InnoDB

2
0
Bronze badge

Re: PostgreSQL

That is not entirely true. MariaDB comes with Aria, a transaction- and crash-safe engine which is completely open-source.

3
0
Anonymous Coward

Re: PostgreSQL

When I first reviewed MySQL as a potential db for a project, I was appalled. It lacked the most fundamental database capabilities, and the real important stuff like RI. I closed the book on MySQL when I discovered these things as well. MySQL was simply not a serious database.

InnoDB came along later and added some transactional capability, but that was a commercial product you had to buy as an add-on IIRC. So a database system which needs transactional capabilities as an add-on WTF?!?!?

Referential integrity was always lost on Monty. He is Swedish, so we can't expect too much.However I recommend you read the drivel in the MySQL manuals from V4 time-frame. Clearly written by someone clueless, a child with a hammer looking for nails, unaware that the screw and screwdriver have been invented!

1
4
WTF?

Say, what?

OK, I've heard strong opinions for and against MySQL that are well thought out and reasonable, but

"He is Swedish, so we can't expect too much."

Seriously?

3
0
Silver badge
Happy

Re: PostgreSQL

> It lacked the most fundamental database capabilities, and the real important stuff like RI.

It depends what you want it for. Perhaps RI isn't important because its used as a content cache. There are some places where speed and low resource requirements are more important than accuracy. UDP, JPG and MPEG4 don't fail as protocols because they are lossy.

Just don't use it for your finance database!

0
0
Silver badge

Re: PostgreSQL

It depends what you want it for. Perhaps RI isn't important because its used as a content cache.

Sure, but then it isn't a relational database but a datastore that supports SQL.

1
0
Meh

Satellite Server

Strange enough, if you operate a Redhat Satellite Server, it solely relies on an Oracle Database; no MySQL, MariaDB, PostgreSQL or something.

0
1

Re: Satellite Server

You can use PostgreSQL if you like - it's a supported option.

3
0
Silver badge
Pint

Re: Satellite Server

There is no obligation to eat your own dogfood .... if you aren't usure about the F/OSS databases, shell out for Miracle Oracle.

0
0
Silver badge
Linux

Re: Satellite Server

Want to bet how long will the Oracle dependency last?

0
0
Anonymous Coward

Re: Satellite Server

No you can't - that's Spacewalk. RHSS requires an Oracle database.

0
0
cmp

Re: Satellite Server

Satellite was first released in 2001, long before Sun bought MySQL let alone the release of MariaDB, so its use of Oracle is a reflection on the state of open source databases 12 years ago. Back then both postgres and mysql were definitly not "enterprise ready" and Red Hat and Oracle were good buddys (it was long before OEL as well) so the choice of using the free (as in beer) edition of oracle seemed like a really good idea.

Things have changed quite a lot since then!

Support for Satellite on Postgres is in the works (Spacewalk already uses it), and I think you can be pretty sure the (rewritten from scratch) Satellite 6 will not have Oracle as an option...

0
0
Silver badge
Go

Long overdue.

RH should have done this over a year ago. I suppose the wait was for support verification. Now, all they need next is an automated tool to migrate MariaDB instances to PostgreSQL instances and the job's done.....

1
2
Silver badge

Re: Long overdue.

RH do not make major changes in their shipped products - ever.

Changing to Maria was NOT an option when RH6 came out and as a result never will be.

I get extremely frustrated with RH's software policies but they have them to keep the big clients happy. If you want agility use something else or import a 3rd party rpm.

6
0
Anonymous Coward

Re: Long overdue.

Major changes bring in major bugs.

The entire point of RH Linux is stability.

0
0
Silver badge

Re: Long overdue.

The entire point of RH Linux is stability.

Don't confuse stasis for stability. The entire point of RH Linux is reassurance: "don't worry if software X is no longer supported by the developers, we will continue to look after it for you…" . RedHat is like the Microsoft of Linux by promising extremely long support cycles for its software. While this doesn't work for me (I prefer the BSD approach of a stable OS and software ports) it obviously does work for a lot of CIOs. Long term the approach is similar to other vendors: dependency by deskilling.

0
0
Thumb Up

another one bites the dust

So what else is new?

- Oracle takes over some open source project by some acquisition

- Lawyers replace developers, threaten to sue for trademark violations etc.

- Developers run away

- Forks happen - in the case of MySQL: both MariaDB and PerconaDB

- Everyone notices that the forks are better

- Users switch to new forks - in this case notables are Wikipedia and now Red Hat

- The Oracle version, encumbered by legalese, restrictive licences, plagued with security issues, terrible package management etc. descends into total irrelevance

- Happiness ensues

16
1
Silver badge
Thumb Up

Re: elreg subscriber Re: another one bites the dust

".....- The Oracle version, encumbered by legalese, restrictive licences, plagued with security issues, terrible package management etc. descends into total irrelevance....." You missed out one part of it all:

- Oracle's salesgrunts enjoy the fractured nature of the My SQL base, allowing them FUD MySQL and forks, to sell more Oracle DB licenses and enjoy the problems caused to what was a competitor product, the real reason behind the Oracle purchase of MySQL.

3
0
Silver badge
Devil

Re: elreg subscriber another one bites the dust

Well... technically, MySQL was bought by Sun at the price of "1 Instagram" before .... the unnamable¹ involving a monstrous presence from Redwood occurred.

[1] ...My constant talk about 'unnamable' and 'unmentionable' things was a very puerile device, quite in keeping with my lowly standing as an author. I was too fond of ending my stories with sights or sounds which paralysed my heroes’ faculties and left them without courage, words, or associations to tell what they had experienced. (H.P. Lovecraft, speaking as "Randolph Carter")

2
0
Anonymous Coward

Re: another one bites the dust

And Oracle succeed in removing an Open Source product from contending with their own DB, by fragmenting it into irrelevance.....

Meanwhile Microsoft SQL Server continues to grow market share.....

1
3
Anonymous Coward

Re: elreg subscriber another one bites the dust

"the real reason behind the Oracle purchase of MySQL."

I see "Matt the Moron" (tm) is at work today.

Oracle did not buy MySQL. They bought Sun Microsystems for quite a considerable sum of money. In the purchased package of assets, was all sorts of detritus, some of it useful and some of it useless.

MySQL was part of the purchase, it was not THE purchase.

Dweeb

4
0

Re: another one bites the dust

You forgot the part where Oracle eventually has to purchase MariaDB, rinse, repeat, start again.

0
0

This post has been deleted by its author

Anonymous Coward

Transparency and maintainability

Another reason which may have motivated RH to switch away from MySQL is the lack of transparency and openness from Oracle: maintaining old versions has become very difficult because Oracle only maintains the HEAD (where vulnerability fixes are drowned with all the new features, which makes it almost impossible for a third-party to isolate the fixes) and doesn't communicate the individual patches.

MySQL may still have, technically, an "open-source" core but this means nothing any more because of the closed project management model Oracle uses. For all intents and purposes it's the same as closed-source software now.

When, last year, a Debian security upgrade had to ship a new upstream version (even though it introduced functional changes, which is normally a no-no for Debian) instead of just backporting, as usual, the fixes to the stable version, then you knew something terribly wrong was happening.

9
0
Silver badge

Re: Transparency and maintainability

I'm a bit surprised that Debian haven't already switched to MariaDB; but then again, they were slow adopting XOrg instead of XFree86, most probably because that was how long it took to test everything still worked. Maybe MariaDB will make it into Jessie + 1 .....

0
0
K
Bronze badge

Long overdue

For to long MySQL has been a drag on Open Source DB development, even before Sun brought it the development and release cycle was way to slow.

But I do disagree with the call for Postgre, reason being I'm sure MySQL has far larger user base that Postgre (that is a personal assumption) and I'm fairly certain Redhat would not want to alienate all those lucrative customers.

Another point that nobody has mentioned, is a couple of years ago Oracle tried to stick it to Redhat by effectively forking the Redhat packages and even offering to sell support for Redhat Enterprise at a cheaper price. Though I doubt this would be a motivator for Redhat to finally show the middle finger ;)

1
0

Re: Long overdue

Moving from any MySQL-based database to PostgreSQL as a default would be good for everyone. PostgreSQL is both awesome in terms of features and an actual real database built on actual real database principles. The more people that use it for their bigger/more important projects the better.

For the smaller just-need-a-structured-datastore needs just use SQLite.

5
0
K
Bronze badge

Re: Long overdue

@SP - What exactly is it that makes Postgre a more "real" database than MySQL and any of its variants?

2
0
Silver badge
Paris Hilton

Re: Long overdue

People ain't quite sure about that:

http://www.wikivs.com/wiki/MySQL_vs_PostgreSQL

0
0
Anonymous Coward

Re: Long overdue

Postgresql is not named after characters in the Moomin books (Yes, I know they are the misses Widenius as well). That makes it more serious. Doesn't it?

0
0
Silver badge

Three-way split

There really isn't much to see here: RHEL is a commercial offering with companies apparently happy to pay RedHat to support software versions ad infinitum; Oracle is pushing ahead with improvements to MySQL (5.6 really does look to be getting quite usable) and happy to favour paying customers over "freeloaders". Both are pursuing vertical integration strategies.

Postgres is filling the niche of a full-fledged RDBMS with no strings attached, with commercial support for those who want it: Enterprise DB has a nice model for companies wanting to get off Oracle and 2nd Quadrant has just announced "platinum" support.

I must admit I've never really understood the value proposition of MyASM beyond its apparent ubiquity and some niche use cases, where speed is valued over integrity at all costs. The ubiquity lowers the barriers to entry for newbies but at the cost of encouraging poor design practices.

1
0

One unique feature

of MyISAM tables is the MERGE table definition. It allows you to merge two tables, even if they are in different databases, so that they look like one table to anything querying the DB. Kinda like a view, but the difference is that you can then specify which table is writeable and all writes will go to that table only.

This can come in very, very handy if you have to do something like have common data in several DBs but can't change the application. I don't know of any way to easily do this in other DBs, although there are some things which come close.

It does have some limitations, like the tables having to be exactly the same and you need to make sure to avoid key collisions. But when your use case calls for something like MERGE, it's a good solution.

2
0
Silver badge

Re: One unique feature

Thanks for the info, I didn't know about it. I'm sure you'll agree it's not a terribly common requirement and not the sort of thing that beginners should be confronted with, unless they pick it up as a practice and start designing for it!

0
0

This post has been deleted by a moderator

Silver badge

Re: Oracle bad will

Sun's "opensource" is usually CDDL, which is legally incompatible with GPL (deliberately so).

This makes it difficult to incorporate things into a single distribution. (There are workarounds, but they're fiddly)

0
0
Anonymous Coward

Re: Oracle bad will

It was great news for Microsoft though - SQL server revenue grew 16% just last year alone...It now has a well estabished and proven lower TCO than any other enterprise ready choice.....

1
4
Anonymous Coward

Oracle alienated open source developers and users alike, and pretty much messed up any open source they acquired or released. At the same time they blatantly fork other open source (can't blame them, GPL allows it), for example RHEL, which became the costly Oracle Linux. Stuff which they previously provided as open source suddenly is no longer supported on other Linux platforms (e.g. OCFS2, which will in not available for RHEL 6, but was for 5.x where they still deliver patches).

Java didn't exactly shine, and I doubt all the holes came from Sun's code. MySQL becomes less usable unless you have too much money to spend; hence forks like MariaDB and Percona gain traction.

Then there was the massacre around OpenSolaris. I'm glad to see that IllumOS, Omni, OpenIndiana, SmartOS and others build on solid SunOS roots and take it from there.

You used to see Oracle developers contributing to many open source projects, but that's becoming less common these days from what I've seen in changelogs (I may be wrong, though).

You get the feeling that Oracle got involved (and acquired) open source *because* they wanted it dead. Not surprising, but still far from nice. (On a side note: many ex Sun developers are working for Oracle's competitors these days.)

I doubt that it drives people to buying Oracle licenses for anything. Rather the opposite. People increasingly often double-check whether or not Oracle is involved with something they'd like to use, and if that's the case, they look for other options.

Mismanagement, as Eadon put it, is indeed what it looks like. In the grand scheme of things and the long-term anyway. It was a short-sighted attempt to remove competition. But Oracle doesn't have enough money to buy all competition now and in the future. I hope. Plus, they underestimated the open source community. They tried hard to kill OpenSolaris, to go back to that example, but it didn't keep excellent developers from carrying on.

I'm waiting for the day where Java becomes proprietary and starts to cost money. Oh well, maybe it would have been already if it wasn't full of holes, which is hard to sell at typical Oracle rates.

5
0

Page:

This topic is closed for new posts.