Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.databases.postgresql > #476 > unrolled thread

why is postgresql better than mariadb?

Back to article view | Back to comp.databases.postgresql

Contents

  why is postgresql better than mariadb? johannes falcone <visphatesjava@gmail.com> - 2013-08-15 20:20 -0700
    Re: why is postgresql better than mariadb? Jasen Betts <jasen@xnet.co.nz> - 2013-08-17 12:54 +0000
    Re: why is postgresql better than mariadb? Harry Tuttle <SOZRBLNTLEEE@spammotel.com> - 2013-08-18 21:40 +0200
      Re: why is postgresql better than mariadb? Mario Splivalo <majk@fly.srk.fer.hr> - 2013-09-12 19:57 +0000
        Re: why is postgresql better than mariadb? Harry Tuttle <SOZRBLNTLEEE@spammotel.com> - 2013-09-12 22:16 +0200
        Re: why is postgresql better than mariadb? Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-06 08:21 +0000
          Re: why is postgresql better than mariadb? Dimitri Fontaine <dimitri@2ndQuadrant.fr> - 2013-10-06 12:10 +0200
            Re: why is postgresql better than mariadb? Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-06 20:44 +0000
              Re: why is postgresql better than mariadb? Dimitri Fontaine <dimitri@2ndQuadrant.fr> - 2013-10-06 23:17 +0200
            Re: why is postgresql better than mariadb? Harry Tuttle <SOZRBLNTLEEE@spammotel.com> - 2013-10-07 18:03 +0200
              Re: why is postgresql better than mariadb? Dimitri Fontaine <dimitri@2ndQuadrant.fr> - 2013-10-07 18:30 +0200
                Re: why is postgresql better than mariadb? Milo Minderbinder <milo@mmenterprises.invalid> - 2013-10-09 01:51 +0000
              Re: why is postgresql better than mariadb Milo Minderbinder <milo@mmenterprises.invalid> - 2013-10-09 01:46 +0000
                Re: why is postgresql better than mariadb Robert Klemme <shortcutter@googlemail.com> - 2013-10-09 08:41 +0200
                  Re: why is postgresql better than mariadb Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-25 03:24 +0000
          Re: why is postgresql better than mariadb? Fredrik Jonson <fredrik@jonson.org> - 2013-10-11 12:17 +0000
            Re: why is postgresql better than mariadb? Paul Sephton <prsephton@gmail.com> - 2013-10-14 03:47 -0700
              Re: why is postgresql better than mariadb? Harry Tuttle <SOZRBLNTLEEE@spammotel.com> - 2013-10-14 18:52 +0200
                Re: why is postgresql better than mariadb? Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-25 03:58 +0000
              Re: why is postgresql better than mariadb? Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-25 03:55 +0000
            Re: why is postgresql better than mariadb? Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-25 03:50 +0000
              Re: why is postgresql better than mariadb? Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-25 04:07 +0000
              Re: why is postgresql better than mariadb? Dimitri Fontaine <dimitri@2ndQuadrant.fr> - 2013-10-25 10:09 +0200
              Re: why is postgresql better than mariadb? johannes falcone <visphatesjava@gmail.com> - 2013-12-07 10:52 -0800
      Re: why is postgresql better than mariadb? johannes falcone <visphatesjava@gmail.com> - 2013-12-07 10:58 -0800
    Re: why is postgresql better than mariadb? Mladen Gogala <gogala.mladen@gmail.com> - 2013-10-06 08:24 +0000
      Re: why is postgresql better than mariadb? johannes falcone <visphatesjava@gmail.com> - 2014-04-09 02:32 -0700
    Re: why is postgresql better than mariadb? Marcello Dias <marcellaodobrasil@gmail.com> - 2013-11-29 07:39 -0800
      Re: why is postgresql better than mariadb? johannes falcone <visphatesjava@gmail.com> - 2014-04-09 02:33 -0700
    Re: why is postgresql better than mariadb? Marcello Dias <marcellaodobrasil@gmail.com> - 2013-11-29 09:17 -0800
      Re: why is postgresql better than mariadb? Robert Klemme <shortcutter@googlemail.com> - 2013-12-02 21:18 +0100
    Re: why is postgresql better than mariadb? Marcello Dias <marcellaodobrasil@gmail.com> - 2013-11-30 05:59 -0800
      Re: why is postgresql better than mariadb? eric@invalid.com (EricF) - 2013-12-04 04:11 +0000
      Re: why is postgresql better than mariadb? johannes falcone <visphatesjava@gmail.com> - 2014-04-09 02:37 -0700
    Re: why is postgresql better than mariadb? Marcello Dias <marcellaodobrasil@gmail.com> - 2013-12-07 02:53 -0800
    Re: why is postgresql better than mariadb? yangnw@live.com - 2014-05-12 08:06 -0700

Page 1 of 2  [1] 2  Next page →

#476 — why is postgresql better than mariadb?

why?

[toc] | [next] | [standalone]

#479

On 2013-08-16, johannes falcone <visphatesjava@gmail.com> wrote:
> why?

How about SQL standards compliance?

probably everything else that's wrong with mysql too (with one
exception)


-- 
⚂⚃ 100% natural

--- news://freenews.netfront.net/ - complaints: news@netfront.net ---

[toc] | [prev] | [next] | [standalone]

#480

johannes falcone wrote on 16.08.2013 05:20:
> why?
>

Things that MySQL (and I believe MariaDB as well) still doesn't have (and PostgreSQL has):

deferrable constraints
check constraints
Non transactional FK evaluation (http://dev.mysql.com/doc/refman/5.5/en/innodb-foreign-key-constraints.html)
   --> constraints are always checked row-by-row, not on statement level
recursive queries
table functions
common table expressions
windowing functions
function based index
partial index
transactional DDL
full text search on transactional tables (changed with 5.6)
GIS features on transactional tables
MINUS or INTERSECT operator
stemming in full text search
user defined aggregate functions
domains or user defined types
statement level triggers
conditional triggers
   Triggers are not fired from cascading deletes
sequences
EXPLAIN for non-select statements or prepared statements (seems to have changed with MySQL 5.6)
No descending indexes
Sub-Selects and derived tables are still 2nd class "citizens"
--> cannot use table being updated in a subselect
--> cannot use derived tables in views
rename columns that are part of a foreign key constraint
call stored procedures in triggers


Things that Postgres doesn't have that MySQL/MariaDB has

WITH CHECK option for views
declarative partitioning

[toc] | [prev] | [next] | [standalone]

#485

On 2013-08-18, Harry Tuttle <SOZRBLNTLEEE@spammotel.com> wrote:
> johannes falcone wrote on 16.08.2013 05:20:
>> why?
>>
>

> Things that Postgres doesn't have that MySQL/MariaDB has
>
> WITH CHECK option for views
> declarative partitioning
>

Master-master replication. I'd like to have that one in PostgreSQL.

	Mario
-- 
"I can do it quick. I can do it cheap. I can do it well. Pick any two."

Mario Splivalo
majk@fly.srk.fer.hr

[toc] | [prev] | [next] | [standalone]

#486

Mario Splivalo wrote on 12.09.2013 21:57:

>> Things that Postgres doesn't have that MySQL/MariaDB has
>>
>> WITH CHECK option for views
>> declarative partitioning
>>
>
> Master-master replication. I'd like to have that one in PostgreSQL.

I was more focused on SQL language features.

And of course you are right, there is no built-in solution for that in PostgreSQL.

But you could get a mulit-master replication working with Slony, Bucardo or pgPool
(although the documentation is not really good on that)

[toc] | [prev] | [next] | [standalone]

#488

On Thu, 12 Sep 2013 19:57:45 +0000, Mario Splivalo wrote:


> Master-master replication. I'd like to have that one in PostgreSQL.
> 
> 	Mario
> --

Also, Mariadb has hints. No hints in PostgreSQL. That in itself is a 
sufficient reason not to use PostgreSQL.



-- 
Mladen Gogala
The Oracle Whisperer
http://mgogala.byethost5.com

[toc] | [prev] | [next] | [standalone]

#490

Mladen Gogala <gogala.mladen@gmail.com> writes:
> Also, Mariadb has hints. No hints in PostgreSQL. That in itself is a 
> sufficient reason not to use PostgreSQL.

No hints means that when a new release is out, you don't have to spend
time fixing all your hinted queries against the new incompatible set of
query hints.

It also means that planner and opimizer bugs are actually fixed as they
as doscovered.

And if you knew how to use PostgreSQL, you would knew it actually has
some kind of hinting due to SQL standard optimiser barriers constructs
and some implementation details.

Then again it's all about use cases. Some use cases might be inadequate
and then PostgreSQL isn't the best tool for the job. Yes, that can
happen.

Regards,
-- 
Dimitri Fontaine
PostgreSQL DBA, Architecte

[toc] | [prev] | [next] | [standalone]

#491

On Sun, 06 Oct 2013 12:10:51 +0200, Dimitri Fontaine wrote:

> No hints means that when a new release is out, you don't have to spend
> time fixing all your hinted queries against the new incompatible set of
> query hints.

It also means that you don't have control over plans produced by the 
optimizer. Since Postgres optimizer has difficulties with things like 
partitioning or correlated columns, that means that it is not possible to 
fix the plan when necessary. The benefit of not having to fix the hints 
does not compensate for not having the control over the generated plans. 
This shortcoming is so big that I migrated one database from PostgreSQL 
to MongoDB, just because of that. When there is one SQL to be fixed 
quickly, because someone is waiting for the application, hints are a 
necessity.

And the "set enable_<method>" command is not a hint. The scope is wrong, 
the scope is the entire session instead of single SQL. Disabling or 
enabling access method when all you want is for one specific SQL 
statement to use one specific index is simply wrong and creates a mess 
for other SQL. That, among other inadequacies of PostgreSQL, is the 
primary reason why PostgreSQL did not catch on. The databases that came 
much later, like MySQL in 3 different flavors or MongoDB wiped the floor 
with PostgreSQL, precisely because of things like that. Funny thing is 
that commercial PostgreSQL versions, like EnterpriseDB or Vertica have 
hints. In other words, you have to pay for hints in the world of Postgres. 
You don't have to pay for them in the world of MySQL. Since hints are so 
indispensable, I will simply neither recommend nor use the database that 
doesn't support them. It's as simple as that.



-- 
Mladen Gogala
The Oracle Whisperer
http://mgogala.byethost5.com

[toc] | [prev] | [next] | [standalone]

#492

Mladen Gogala <gogala.mladen@gmail.com> writes:
>  Since hints are so 
> indispensable, I will simply neither recommend nor use the database that 
> doesn't support them. It's as simple as that.

I'm happy to know that you're happy with your choices and trade-offs.
Mine are different.

Regards,
-- 
Dimitri Fontaine
PostgreSQL DBA, Architecte

[toc] | [prev] | [next] | [standalone]

#493

Dimitri Fontaine wrote on 06.10.2013 12:10:
> No hints means that when a new release is out, you don't have to spend
> time fixing all your hinted queries against the new incompatible set of
> query hints.
>
> It also means that planner and opimizer bugs are actually fixed as they
> as doscovered.
>
> And if you knew how to use PostgreSQL, you would knew it actually has
> some kind of hinting due to SQL standard optimiser barriers constructs
> and some implementation details.
>
> Then again it's all about use cases. Some use cases might be inadequate
> and then PostgreSQL isn't the best tool for the job. Yes, that can
> happen.

Don't feed the troll.

Mladen is out on a holy war to bash Postgres wherever he can because he takes
that fact that Postgres works differently than Oracle as a personal offense.

The fact that hints are apparently the only way he knows to tune a statement, speaks for itself.

[toc] | [prev] | [next] | [standalone]

#494

Harry Tuttle <SOZRBLNTLEEE@spammotel.com> writes:
> Mladen is out on a holy war to bash Postgres wherever he can because he takes
> that fact that Postgres works differently than Oracle as a personal offense.

Thanks for that context, I'll know not to answer next time!

-- 
Dimitri Fontaine
PostgreSQL DBA, Architecte

[toc] | [prev] | [next] | [standalone]

#496

On Mon, 07 Oct 2013 18:30:52 +0200, Dimitri Fontaine wrote:

> Harry Tuttle <SOZRBLNTLEEE@spammotel.com> writes:
>> Mladen is out on a holy war to bash Postgres wherever he can because he
>> takes that fact that Postgres works differently than Oracle as a
>> personal offense.
> 
> Thanks for that context, I'll know not to answer next time!

There is nothing to say. Any doubters should only take a look at the
following page:

http://it.toolbox.com/blogs/database-soup/why-postgresql-doesnt-have-query-hints-44121?rss=1

and be aware that the author is on PostgreSQL design committee to know all
there is to know about PostgreSQL. And I don't really care whether you will
reply or not. My warning is not meant for the PostgreSQL zealots, it's 
directed toward anyone who may consider using it. Don't. It's as simple as
that.



-- 
I love the smell of napalm in the morning. It smells like....victory.
(Apocalypse Now)

[toc] | [prev] | [next] | [standalone]

#495 — Re: why is postgresql better than mariadb

On Mon, 07 Oct 2013 18:03:49 +0200, Harry Tuttle wrote:

> Mladen is out on a holy war to bash Postgres wherever he can because he 
takes
> that fact that Postgres works differently than Oracle as a personal 
offense.

I'm not on waging a war, I am just warning people of deception. Postgres 
is a bait & switch thing. I have wasted approximately a year of my life 
trying to learn it. I'm just pointing out that Postgres is an inadequate DB
for the commercial use. Simply put, without the missing features, Postgres 
is inadequate for any serious company to entrust data to it.
There is nothing wrong with not being similar to Oracle. DB2 and SQL 
Server are databases which are not similar to Oracle, yet they both have
decent partitioning, built-in parallelism and hints. Postgres is simply not
on the same level as those guys. For that matter, neither is MySQL in any 
of its flavours, but MySQL is good enough, precisely because of hints.
In addition to that, MySQL developers are open minded. If there are 
requirements for certain feature, MySQL guys will implement it, while 
PostgreSQL developers is not. There are funny coincidences that are worth 
noting, too. For instance, the guy who is basically the head of design 
committee works for EnterpriseDB. EnterpriseDB is selling a version of 
Postgres with hints. So, in the world of Postgres one has to pay for 
hints. All other databases give them away. That's enough for me. I steer 
clear of Postgres. 
And don't compare Postgres to Oracle. There is no comparison. Stonebraker 
has lost to Ellison before. The fact that PostgreSQL share is not growing 
while Oracle share is growing, despite the atrocious license costs, ought 
to tell you something. If a free database cannot grow when pitched against 
an expensive commercial database, then there is a problem with the freebie.



-- 
I love the smell of napalm in the morning. It smells like....victory.
(Apocalypse Now)

[toc] | [prev] | [next] | [standalone]

#497 — Re: why is postgresql better than mariadb

On 10/09/2013 03:46 AM, Milo Minderbinder wrote:
> On Mon, 07 Oct 2013 18:03:49 +0200, Harry Tuttle wrote:
>
>> Mladen is out on a holy war to bash Postgres wherever he can because he
> takes
>> that fact that Postgres works differently than Oracle as a personal
> offense.
>
> I'm not on waging a war, I am just warning people of deception.

Interesting that you use deception to warn of deception - or how would 
you call using different identities in the same public forum?

	robert

[toc] | [prev] | [next] | [standalone]

#502 — Re: why is postgresql better than mariadb

On Wed, 09 Oct 2013 08:41:25 +0200, Robert Klemme wrote:

> On 10/09/2013 03:46 AM, Milo Minderbinder wrote:
>> On Mon, 07 Oct 2013 18:03:49 +0200, Harry Tuttle wrote:
>>
>>> Mladen is out on a holy war to bash Postgres wherever he can because
>>> he
>> takes
>>> that fact that Postgres works differently than Oracle as a personal
>> offense.
>>
>> I'm not on waging a war, I am just warning people of deception.
> 
> Interesting that you use deception to warn of deception - or how would
> you call using different identities in the same public forum?
> 
> 	robert

It wasn't a deception, I was using a different PC. It was  completely 
accidental, not by design. 



-- 
Mladen Gogala
The Oracle Whisperer
http://mgogala.byethost5.com

[toc] | [prev] | [next] | [standalone]

#499

In <pan.2013.10.06.08.21.30@gmail.com> Mladen Gogala wrote:

>  Also, Mariadb has hints. No hints in PostgreSQL. That in itself is a 
>  sufficient reason not to use PostgreSQL.

I don't know what strange database environments you're the DBA or designer
of Mladen, but your dependency to a fault on query hints is really curious.

You can easily use Postgresql, Mysql, Mariadb and many other database
engines for decades in regular service with various interesting loads and
volume requirements without ever having a problem that requires hints to be
resolved.

I'm sure you'd be happy to provide us with contrieved examples of how hints
saves your day on regular basis. However there is ample proof that
Postgresql works ever so well as other database systems, even though it
doesn't implement hints. If there wasn't, don't you think developers and
system designers would abandone Postgresql in droves? Hint: They don't.

For anyone who want to know why the Postgresql community currently do not
consider hints to be very interesting, I encourage you to read this:

http://wiki.postgresql.org/wiki/OptimizerHintsDiscussion

-- 
Fredrik Jonson

[toc] | [prev] | [next] | [standalone]

#500

<snip>
> ... I encourage you to read this:
> 
> http://wiki.postgresql.org/wiki/OptimizerHintsDiscussion
> 


MLaden sounds very much an example of point 3 in the list of reasons against hints:
 - Encouraging bad DBA habits slap a hint on instead of figuring out the real issue. 

[toc] | [prev] | [next] | [standalone]

#501

Paul Sephton wrote on 14.10.2013 12:47:
> <snip>
>> ... I encourage you to read this:
>>
>> http://wiki.postgresql.org/wiki/OptimizerHintsDiscussion
>>
>
>
> MLaden sounds very much an example of point 3 in the list of reasons against hints:
>   - Encouraging bad DBA habits slap a hint on instead of figuring out the real issue.
>

That discussion actually reminds me of a discussion I once had with a SQL Server user that migrated to
Oracle and who as claiming that Oracle was totally useless because it does not have a NOLOCK hint and that
would be the only way to improve the speed of queries in a high transaction system.

[toc] | [prev] | [next] | [standalone]

#505

On Mon, 14 Oct 2013 18:52:13 +0200, Harry Tuttle wrote:

> That discussion actually reminds me of a discussion I once had with a
> SQL Server user that migrated to Oracle and who as claiming that Oracle
> was totally useless because it does not have a NOLOCK hint and that
> would be the only way to improve the speed of queries in a high
> transaction system.

How come? Would you care to explain a bit? The user you are talking about 
doesn't understand Oracle and is asking for something that Oracle doesn't 
need. Are you claiming that PostgreSQL optimizer is so good that it 
doesn't need hints? Did I understand you correctly?



-- 
Mladen Gogala
The Oracle Whisperer
http://mgogala.byethost5.com

[toc] | [prev] | [next] | [standalone]

#504

On Mon, 14 Oct 2013 03:47:44 -0700, Paul Sephton wrote:


> MLaden sounds very much an example of point 3 in the list of reasons
> against hints:
>  - Encouraging bad DBA habits slap a hint on instead of figuring out the
>  real issue.

I would be interested to know your qualifications as a DBA, when talking 
about "bad DBA habits"? The "real issue" is always under-performing SQL 
which needs to be improved. Sometimes, it is possible to do quickly, 
sometimes it isn't. I can see that you've never had a VP staring over 
your shoulder and asking "is it fixed yet"? I have lived through such 
situations. And they can only be resolved by hints.
Also, how in the world do you propose to deal with correlated columns and 
partitioned tables? PostgreSQL optimizer is atrocious when it comes to 
partitioned tables. 



-- 
Mladen Gogala
The Oracle Whisperer
http://mgogala.byethost5.com

[toc] | [prev] | [next] | [standalone]

Page 1 of 2  [1] 2  Next page →

Back to top | Article view | comp.databases.postgresql

csiph-web