Обсуждение: Is PostGreSql's Data storage mechanism "inferior"?

Hi all,

I'm new to PostGreSql.

http://searchyourwebhost.com/web-hosting/articles/insight-database-hosting-using-sql

Check out the link. I am starting out on a new personal project & had zeroed in on PostGreSql with Mono-ASP.NET as ideal for my needs, mainly owing to a PostGreSql whitepaper.
Now, I chanced upon the article above. I've pasted the cons as mentioned in the article, and would like the community feedback on it, especially with regards to the "inferior Data Storage mechanism".

The cons of PostgreSql Hosting
* Performance considerations: Inserts and Updates into the PostgreSql database is much slower compared to MySql. PostgreSql hosting thus might slow down the display of the web page online.
* BSD license issues: Since PostgreSql comes under the Berkeley license scheme, this is again considered to be too open.
* Availability of inferior Data Storage mechanism: PostgreSql uses Postgres storage system, which is not considered to be transaction sae during PostgreSql hosting.
* Its not far-flung: While MySql hosting and MSSql hosting have deeply penetrated into the market, PostgreSql hosting still remains to be passive in the database hosting market.
* Non-availability of required assistance for PostgreSql hosting: Assistance is being provided via mailing lists. However there is no guarantee that the issue faced during PostgreSql hosting would be resolved.

Thanks!

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Wed, 30 Jan 2008 15:11:05 -0800
"Swaminathan Saikumar" <swami@giveexam.com> wrote:

> Hi all,
> 
> I'm new to PostGreSql.
> 
> http://searchyourwebhost.com/web-hosting/articles/insight-database-hosting-using-sql

> The cons of PostgreSql Hosting
> * Performance considerations: Inserts and Updates into the PostgreSql
> database is much slower compared to MySql. PostgreSql hosting thus
> might slow down the display of the web page online.

Not when data integrity comes into play.

> * BSD license issues: Since PostgreSql comes under the Berkeley
> license scheme, this is again considered to be too open.

Exactly how is too open a bad thing?


> * Availability of inferior Data Storage mechanism: PostgreSql uses
> Postgres storage system, which is not considered to be transaction
> sae during PostgreSql hosting.

Uhmm that is a blatant lie.

> * Its not far-flung: While MySql hosting and MSSql hosting have deeply
> penetrated into the market, PostgreSql hosting still remains to be
> passive in the database hosting market.

I know many postgresql hosting companies.

> * Non-availability of required assistance for PostgreSql hosting:
> Assistance is being provided via mailing lists. However there is no
> guarantee that the issue faced during PostgreSql hosting would be
> resolved.

Say what?

http://www.postgresql.org/support/professional_support

Sincerely,

Joshua D. Drake


> 
> Thanks!


- -- 
The PostgreSQL Company since 1997: http://www.commandprompt.com/ 
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFHoREhATb/zqfZUUQRAkRpAJ9gdQAwmWsXPNlut0DJ8/mNgzmytQCdEd8M
0kBugrvVLkPSH4VWBtKYUUE=
=vqnc
-----END PGP SIGNATURE-----

--- On Wed, 1/30/08, Swaminathan Saikumar <swami@giveexam.com> wrote:

> The cons of PostgreSql Hosting
> * Performance considerations: Inserts and Updates into the
> PostgreSql database is much slower compared to MySql. PostgreSql
> hosting thus might slow down the display of the web page online.

" ... might slow down ..."  I guess they didn't check to know one way or ther other whether It does slow down a web
page. The real answer depends upon your they kind of uses you have. 

> * BSD license issues: Since PostgreSql comes under the
> Berkeley license scheme, this is again considered to be too open.

What does "too open" mean?  Is too open a bad thing?

> * Availability of inferior Data Storage mechanism:
> PostgreSql uses Postgres storage system, which is not considered to be transaction sae during PostgreSql hosting.

It seems they misspelled "safe".  But in either case they also seemed to have confused MySQL with PostgreSQL on this
point. Being transactionally "safe" is one of PostgreSQL's strongest points. 

> * Its not far-flung: While MySql hosting and MSSql hosting have deeply
> penetrated into the market, PostgreSql hosting still remains to be passive in the database hosting market.

My gut says that this is probably true. But there are very good hosting sites that provide PostgreSQL.


> * Non-availability of required assistance for PostgreSQL hosting: Assistance is being provided via mailing lists.
Howeverthere is no guarantee that the issue faced during PostgreSQL hosting would be resolved.  

If you have a highly critical application that requires guaranteed assistance you are more than able to purchase it
fromsome top notch consultant companies that support PostgreSQL.  Just check out the PostgreSQL home page.  So I don't
thisthis point is correct either. 

If you don't get too much feed back on this subject, just remember that topics like this come up frequently to the
pointof list member exhaustion.   You can find such discussions if you search the list archive. 

Regards,
Richard Broersma Jr.

Swaminathan Saikumar wrote:
> Hi all,
>
> I'm new to PostGreSql.
>
> http://searchyourwebhost.com/web-hosting/articles/insight-database-hosting-using-sql
What a wonderful article - it's almost worth keeping a copy. It's so bad
it's difficult to know where to start. I think my favourite has to be :
+ MSSql being massive is considered to have the maximum scalability for
database hosting.

And I just love the comment on both MySQL and Postgres about GPL and BSD
being 'too open' - meaning?

Pete
>
> Check out the link. I am starting out on a new personal project & had
> zeroed in on PostGreSql with Mono-ASP.NET as ideal for my needs,
> mainly owing to a PostGreSql whitepaper.
> Now, I chanced upon the article above. I've pasted the cons as
> mentioned in the article, and would like the community feedback on it,
> especially with regards to the "inferior Data Storage mechanism".
>
> The cons of PostgreSql Hosting
> * Performance considerations: Inserts and Updates into the PostgreSql
> database is much slower compared to MySql. PostgreSql hosting thus
> might slow down the display of the web page online.
> * BSD license issues: Since PostgreSql comes under the Berkeley
> license scheme, this is again considered to be too open.
> * Availability of inferior Data Storage mechanism: PostgreSql uses
> Postgres storage system, which is not considered to be transaction sae
> during PostgreSql hosting.
> * Its not far-flung: While MySql hosting and MSSql hosting have deeply
> penetrated into the market, PostgreSql hosting still remains to be
> passive in the database hosting market.
> * Non-availability of required assistance for PostgreSql hosting:
> Assistance is being provided via mailing lists. However there is no
> guarantee that the issue faced during PostgreSql hosting would be
> resolved.
>
> Thanks!

On Jan 30, 2008 5:11 PM, Swaminathan Saikumar <swami@giveexam.com> wrote:
> Hi all,
>
> I'm new to PostGreSql.
>
> http://searchyourwebhost.com/web-hosting/articles/insight-database-hosting-using-sql

I just skimmed through that page and honestly, it's wrong on LOTS of
counts, again and again. For instance, about mysql it has these two
contradicting points:
QUOTE:
Pros:
Availability of Superior Data Storage mechanism: Though prior versions
of MySql supported ISAM/MyISAM mechanisms, later versions starting
from 4.x have started using the InnoDB mechanism, which is considered
to be transaction safe for database hosting.

Cons:
Presence of Inferior Data Integrity mechanism: Though MySql is ACID
(Atomic, Consistent, Isolated, Durable) complaint, when dealing with
deadlocks in database hosting, it uses row-level locking which is
considered to be inferior when compared to Multi Version Concurrency
Control (MVCC).
UNQUOTE:

OK, while I hardly consider InnoDB to be superior to PostgreSQL's
storage engine, it uses MVCC for its locking mechanism!  But, MySQL
really isn't capable of true ACID compliance because it as a whole
doesn't support check constraints, but it accepts the syntax without
actually implementing the feature.

I read one, and it seems like many of the entries don't even make
sense.  This one for MSSQL for instance:

QUOTE:
Pros:
Remarkable Reliability: MSSQL hosting offers high reliability by
having a data manager for reading and writing data to the database.
Even if the client machine crashes, the read and write is not
committed in the database by the data manager. The transaction logs
also help in rollbacks thus paving way to commendable reliability in
MSSql hosting.
UNQUOTE:

WHAT?  What does that mean?  And how does it imply superior
reliability?  I can't see any of that meaning MSSQL stays up longer
than any other database.

Seriously, if I were interviewing someone for a db job, and they
pointed out that they wrote that page I would politely decline to hire
them.

On Jan 30, 2008, at 5:11 PM, Swaminathan Saikumar wrote:

> Hi all,
>
> I'm new to PostGreSql.
>
> http://searchyourwebhost.com/web-hosting/articles/insight-database-
> hosting-using-sql
>
> Check out the link. I am starting out on a new personal project &
> had zeroed in on PostGreSql with Mono-ASP.NET as ideal for my
> needs, mainly owing to a PostGreSql whitepaper.
> Now, I chanced upon the article above. I've pasted the cons as
> mentioned in the article, and would like the community feedback on
> it, especially with regards to the "inferior Data Storage mechanism".
>
> The cons of PostgreSql Hosting
> * Performance considerations: Inserts and Updates into the
> PostgreSql database is much slower compared to MySql. PostgreSql
> hosting thus might slow down the display of the web page online.

Not for inserts.  For updates, nder some workloads, possibly.  A
typical website run on a hosted server, not likely.  While you might
be able to clock single instances of these operations on each of
those databases against each other (and, I emphasize *might*) and
have MySQL come out on top, MySQL is the often demonstrated loser
when you want to scale out and process hundreds to thousands at
once.  Isn't that what you're shooting for with a web app?
>
> * BSD license issues: Since PostgreSql comes under the Berkeley
> license scheme, this is again considered to be too open.

What does that even mean?
>
> * Availability of inferior Data Storage mechanism: PostgreSql uses
> Postgres storage system, which is not considered to be transaction
> sae during PostgreSql hosting.

What I *think* they're getting at there is pure nonsense.  In fact,
the sentence itself is nonsensical.
>
> * Its not far-flung: While MySql hosting and MSSql hosting have
> deeply penetrated into the market, PostgreSql hosting still remains
> to be passive in the database hosting market.

While I'll admit that MySQL hosting is more widespread, calling
Postgres hosting "passive" has no meaning whatsoever.
>
> * Non-availability of required assistance for PostgreSql hosting:
> Assistance is being provided via mailing lists. However there is no
> guarantee that the issue faced during PostgreSql hosting would be
> resolved.

You pay for what you get, i.e there are numerous commercial companies
that offer both paid support and consulting.  For free, as mentioned,
you have the lists which are, incidentally, much better that the free
options available to most other technologies.  Watch this list for a
bit, hardly a week goes by without at least one or two people
professing the support and general help gleaned from it's
participants as their primary decision to go with Postgres.

Basically, the person who made that list pulled a bunch of bullet
points out of the FUD-mosphere that barely make sense as English
sentences and offered them up without a shred of evidence or reference.

Erik Jones

DBA | Emma®
erik@myemma.com
800.595.4401 or 615.292.5888
615.292.0777 (fax)

Emma helps organizations everywhere communicate & market in style.
Visit us online at http://www.myemma.com

On Jan 30, 2008, at 6:22 PM, Richard Broersma Jr wrote:

> If you don't get too much feed back on this subject, just remember
> that topics like this come up frequently to the point of list
> member exhaustion.   You can find such discussions if you search
> the list archive.

Too true.  There's only so many times people can be confronted with
"Defend yourselves!" before they start ignoring it.  On the flip
side, when you approach with "Tell me, what advantages does Postgres
have to offer?" you'll find many people all too willing to step up
with pride.

Erik Jones

DBA | Emma®
erik@myemma.com
800.595.4401 or 615.292.5888
615.292.0777 (fax)

Emma helps organizations everywhere communicate & market in style.
Visit us online at http://www.myemma.com

Thanks everyone. After all the good things I heard about Postgres, I was surprised to see this article; and the point on storage concerned me.
I am glad to see that the article was wrong, not only on the storage engine count, but also on others.
Thanks for the feedback.

Swaminathan Saikumar wrote:
> http://searchyourwebhost.com/web-hosting/articles/insight-database-hosting-using-sql
>
> Check out the link. I am starting out on a new personal project & had
> zeroed in on PostGreSql with Mono-ASP.NET as ideal for my needs,
> mainly owing to a PostGreSql whitepaper.
> Now, I chanced upon the article above. I've pasted the cons as
> mentioned in the article, and would like the community feedback on it,
> especially with regards to the "inferior Data Storage mechanism".
>
> The cons of PostgreSql Hosting
> * Performance considerations: Inserts and Updates into the PostgreSql
> database is much slower compared to MySql. PostgreSql hosting thus
> might slow down the display of the web page online.
> * BSD license issues: Since PostgreSql comes under the Berkeley
> license scheme, this is again considered to be too open.
> * Availability of inferior Data Storage mechanism: PostgreSql uses
> Postgres storage system, which is not considered to be transaction sae
> during PostgreSql hosting.
> * Its not far-flung: While MySql hosting and MSSql hosting have deeply
> penetrated into the market, PostgreSql hosting still remains to be
> passive in the database hosting market.
> * Non-availability of required assistance for PostgreSql hosting:
> Assistance is being provided via mailing lists. However there is no
> guarantee that the issue faced during PostgreSql hosting would be
> resolved.
>
Those "cons" are seriously out of date.
They apply to "very" old versions of PostgreSQL and even that is
stretching it.

The part about the BSD license is bogus. A BSD license is the most
desirable of any Open Source license and gives you the right to use
PostgreSQL in your commercial apps without worry.

The part about  inferior Data Storage mechanism is also flat out wrong.
PostgreSQL uses a MVC system same as Oracle and it also has transaction
logs,PITR etc.

Inserts and updates will be slightly slower than a Non MVC system, but
the human eye would not detect any difference in a web page displaying,
we are talking about miliseconds.  Who does massive amounts of inserts
and updates from a web page anyway?  I have CMS such as Drupal running
on both PostgreSQL and MySQL and I can't tell the difference in the
speed the pages render.

The availability of assistance is also bogus as there are many ways to
get support if you need it including commercial support companies and
Enterprise DB, The mailing list is also very active and effective.

The only part that has any truth to it is the far flung part, and MySQL
is king there, it even dwarfs M$ SQL server.  M$ SQL server is severely
limited for hosting as well since it ONLY runs on windows and most
hosting providers run some form of Unix where M$ cannot play at all.

So in conclusion I would not pay attention to this article, it was
written by someone who really does not know what they are talking about.


Later,

Tony Caduto
AM Software Design
Home of Lightning Admin for PostgreSQL and MySQL

On 01/02/2008, Tony Caduto <tony_caduto@amsoftwaredesign.com> wrote:
> The part about the BSD license is bogus. A BSD license is the most
> desirable of any Open Source license and gives you the right to use
> PostgreSQL in your commercial apps without worry.
While I'm a big fan of the BSD license (for varied reasons) I think that
OpenSource hardliners like RMS would argue that the BSD license is *NOT*
in the true spirit of OpenSource *BECAUSE* of what you list as a bonus
of it ... the locking down of benefits reaped from OpenSource not getting
back into the stream.


Cheers,
Andrej


--
Please don't top post, and don't use HTML e-Mail :}  Make your quotes concise.

http://www.american.edu/econ/notes/htmlmail.htm

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Fri, 1 Feb 2008 10:04:53 +1300
"Andrej Ricnik-Bay" <andrej.groups@gmail.com> wrote:

> On 01/02/2008, Tony Caduto <tony_caduto@amsoftwaredesign.com> wrote:
> > The part about the BSD license is bogus. A BSD license is the most
> > desirable of any Open Source license and gives you the right to use
> > PostgreSQL in your commercial apps without worry.

> While I'm a big fan of the BSD license (for varied reasons) I think
> that OpenSource hardliners like RMS would argue that the BSD license
> is *NOT* in the true spirit of OpenSource *BECAUSE* of what you list
> as a bonus of it ... the locking down of benefits reaped from
> OpenSource not getting back into the stream.

RMS wouldn't make any such argument at all. His argument would be
centered around FREE not OpenSource software.

Joshua D. Drake

> 
> 
> Cheers,
> Andrej
> 
> 


- -- 
The PostgreSQL Company since 1997: http://www.commandprompt.com/ 
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFHonjaATb/zqfZUUQRAjfrAKCZu2KZigaCrNT6c9nbuAFYImRhdQCeI4uT
2gdMn7CA9XExIynw5mFogBs=
=pqTR
-----END PGP SIGNATURE-----

"Andrej Ricnik-Bay" <andrej.groups@gmail.com> writes:
> On 01/02/2008, Tony Caduto <tony_caduto@amsoftwaredesign.com> wrote:
>> The part about the BSD license is bogus. A BSD license is the most
>> desirable of any Open Source license and gives you the right to use
>> PostgreSQL in your commercial apps without worry.

> While I'm a big fan of the BSD license (for varied reasons) I think that
> OpenSource hardliners like RMS would argue that the BSD license is *NOT*
> in the true spirit of OpenSource *BECAUSE* of what you list as a bonus
> of it ... the locking down of benefits reaped from OpenSource not getting
> back into the stream.

The quoted article knocked *both* GPL and BSD as being "too open".  Too
open for whom, he didn't say.  The rest of the article is at about the
same quality level :-(  I have seldom seen such a sterling example of
cluelessness combined with FUD-spouting.

            regards, tom lane

andrej.groups@gmail.com ("Andrej Ricnik-Bay") writes:
> On 01/02/2008, Tony Caduto <tony_caduto@amsoftwaredesign.com> wrote:
>> The part about the BSD license is bogus. A BSD license is the most
>> desirable of any Open Source license and gives you the right to use
>> PostgreSQL in your commercial apps without worry.

> While I'm a big fan of the BSD license (for varied reasons) I think that
> OpenSource hardliners like RMS would argue that the BSD license is *NOT*
> in the true spirit of OpenSource *BECAUSE* of what you list as a bonus
> of it ... the locking down of benefits reaped from OpenSource not getting
> back into the stream.

RMS wouldn't argue that, because he is uninterested in "OpenSource."
That's a term created/popularized by Eric Raymond, who is no friend of
RMS.

RMS is interested in "free software," and considers various
"BSD-related" licenses to be reasonable choices for free software.

   <http://www.gnu.org/philosophy/license-list.html>

He'd prefer that people use the GPL, but I don't think there's
anything overly "hard line" about the notion of the author of a
license preferring it to others.  It would seem mighty odd if he said
something like "I wrote the GPL, but think you should use the <Foo
License> instead."
--
"cbbrowne","@","cbbrowne.com"
http://cbbrowne.com/info/linuxxian.html
Jumping  off a  cliff doesn't  kill you!  It's only  when you  hit the
ground...

On Wed, 30 Jan 2008, Swaminathan Saikumar wrote:

> * Performance considerations: Inserts and Updates into the PostgreSql
> database is much slower compared to MySql. PostgreSql hosting thus might
> slow down the display of the web page online.
> * Availability of inferior Data Storage mechanism: PostgreSql uses Postgres
> storage system, which is not considered to be transaction sae during
> PostgreSql hosting.

As many others have already pointed out that entire site is full of
incorrect information.  There are some potential performance issues on
insert/update in PostgreSQL because the default (and only recommended)
configuration makes sure that new transactions are committed to disk
properly when they end.  This can limit an individual client from being
able to make more than 100 commits/second on typical PC hardware.  But any
other database that is configured for a similar level of robustness has
the same problem.

You can find a more useful comparison of PostgreSQL and MySQL at
http://www.postgresql.org/docs/techdocs.83

--
* Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD