Обсуждение: benchmarks

hi all,

lately at work there has been a debate over
mysql versus postgres

im just looking for independent benchmarks

i personally love postgres 
at work they like mysql

currently we are investigating other possible db solutions

and they are looking at oracle, i think we could save a lot of dollarsz 
if we decided to go to postgres


i was wondering if anyone can share links to  any current independent
benchmarks 

as i would like some real data on these 

or at the very least give me a how to so i can do my own testing!

thanks!

> im just looking for independent benchmarks
Even  if  there  exist  independent   benchmarks,  there  are  none  being
applicable to real life.

> i personally love postgres
> at work they like mysql
>
> currently we are investigating other possible db solutions
Uh? You _love_ this and they  _like_ that? Since when, I dareask, personal
preferences are considered  when chosing a product to  trust corporate data
to? How 'bout using just what's the right tool?

> or at the very least give me a how to so i can do my own testing!
Implement your system in MySQL. Then re-do it from scratch in PG. Compare.
If you still have time and funds, make your choice.


--
contaminated fish and microchips huge supertankers on Arabian tripsoily propaganda from the leaders' lips all about the
futurethere'speople over here, people over there everybody's looking for a little more aircrossing all the borders just
totake their share planning for the future
 
Rainbow, Difficult to Cure

Frankyl, Clayton:
Comparing PostgreSQL to MySQL is like comparing an 18-wheel Kenworth
to a Porsche.  The two are not equivalent ... if you want a simple, very
very fast READ-ONLY database, use MySQL.  If you want a full-featured
transaction-environment database for a huge, complex set of data, use
PostgreSQL.
Any benchmarks you find are likely comapring the two for hosting web
sites, which is unfair to PostgreSQL.  If everything your database needs
to do is serve up page content, go for MySQL.  If you've got to build a
60-user inventory management system, MySQL won't even bring you close -
heck, it's not even close to SQL92-compliant.
Oracle is another game altogether, though I can imagine Postgres
catching up in a few years.  Still, the price tag for Oracle weeds out
all but the very serious and deep-pocketed.  And stay away from MS SQL
Server ... I run two of the damn machines, and they're nothing but
grief.
                -Josh
-- 
______AGLIO DATABASE SOLUTIONS___________________________                                       Josh Berkus  Complete
informationtechnology      josh@agliodbs.com   and data management solutions       (415) 436-9166  for law firms, small
businesses      fax  436-0137   and non-profit organizations.       pager 338-4078                               San
Francisco

>  Comparing PostgreSQL to MySQL is  like comparing an 18-wheel Kenworth to
> a Porsche. The two are not equivalent
That's it. Major differences in features makes any benchmarking `apples to
oranges.'

> ... if you want a simple, very
> very fast READ-ONLY database, use MySQL.
BTW, can it  be used on a  r/o partition, or it still  writes something to
the files (locks, stats)?

>  If you want a full-featured transaction-environment database for a huge,
> complex set of data, use PostgreSQL.
Even if the  system is relatively small and simple....  Everything will go
down in fires, this is the law. So,  the first thing to think about is what
will  your options  be when  you're trying  to keep  the parts.  Stress and
torture tests would be more usefull than benchmarking.

>  Any benchmarks  you find are  likely comapring  the two for  hosting web
> sites, which is unfair to PostgreSQL.
IIRC,  there was  one  in favor  PGSQL  (wasn't it  run  by the  Bridge?).
However, the very  fact of benchmarking these two  along with non-disclosed
commercial products, where  PGSQL wins with flying colours,  IMHO does more
harm than good.

> If everything your database needs
> to do is serve up page content, go for MySQL.
How 'bout flat files? They will be even faster. Main advantage of DBMS, as
I  see  it, is  a  possibility  to  build searchable,  logically  connected
structures, with consistency and concurrency  supported on the server side.
With MySQL  one has neither  referential integrity, nor server  side logic.
Moreof, HTML pages are bad as  textual attributes even when tuple/page size
limitation can be easily worked around.If you  want dynamic  content, SSI  and CGI will  do, and  there's nothing
about DBMS. Storing page templates as BLOBs gives nothing but slowdown.

>  If  you've  got   to  build  a  60-user   inventory  management  system,
> MySQL  won't  even  bring you  close  -  heck,  it's  not even  close  to
> SQL92-compliant.
It's still  possible, one  can even  get it for  $ 200,  but TCO  will sky
rocket when it starts to phreak, or when business rules will be changed.


--
contaminated fish and microchips huge supertankers on Arabian tripsoily propaganda from the leaders' lips all about the
futurethere'speople over here, people over there everybody's looking for a little more aircrossing all the borders just
totake their share planning for the future
 
Rainbow, Difficult to Cure

As a former Oracle developer, I decided to start working with cheaper
DBMS's. After a quick look on the market, PostgreSQL was the only one really
worth looking into.

But people liked MySQL, and I had to look for benchmarks. I found only one
(attached).

Not satisfied, I got PostgreSQL and MySQL, compiled and installed both.
MySQL comes with bechmarking tools, so I decided to use them. Very
impressive results for MySQL, obviously.

For PgSQL, the problem was that the benchmarks were not optimized. Not even
bulk loading was used. So I optimized it, and ran the test with PgSQL "NO
FSYNC" option.

The results were that PgSQL was slower than MySQL only by a factor of 2 or 3
(say, 3 seconds for MySQL against 6 or 8 seconds for PgSQL). Pretty good in
my opinion.

Note that for READ-ONLY access, PgSQL is practically as fast as MySQL. And,
according to the attached document (not written by me), PgSQL gets faster
when the SELECT involves several joined tables.

> -----Original Message-----
> hi all,
>
> lately at work there has been a debate over
> mysql versus postgres
>
> im just looking for independent benchmarks
>
> i personally love postgres
> at work they like mysql
>
> currently we are investigating other possible db solutions
>
> and they are looking at oracle, i think we could save a lot of dollarsz
> if we decided to go to postgres
>
>
> i was wondering if anyone can share links to  any current independent
> benchmarks
>
> as i would like some real data on these
>
> or at the very least give me a how to so i can do my own testing!

MTcW: 

Pick the database which allows your programmers
to get the job done. If the system is too slow, find out
if there are ways you could speed it up, and then if that
doesn't make you happy, get a faster server.

In my opinion it's not worth making the programmer's life
more difficult to go with a database which is difficult to
work with.

Also, make sure the database you choose is reliable.
One of the mySQL, mSQL (I confuse the two) servers has
the problem that they lock the entire db when working, 
so only one client can connect at a time. Seems like
a problem to me.


Troy


> 
> 
> hi all,
> 
> lately at work there has been a debate over
> mysql versus postgres
> 
> im just looking for independent benchmarks
> 
> i personally love postgres 
> at work they like mysql
> 
> currently we are investigating other possible db solutions
> 
> and they are looking at oracle, i think we could save a lot of dollarsz 
> if we decided to go to postgres
> 
> 
> i was wondering if anyone can share links to  any current independent
> benchmarks 
> 
> as i would like some real data on these 
> 
> or at the very least give me a how to so i can do my own testing!
> 
> thanks!
> 
> 
> 
> 
> 
>