Tom,

Perhaps its because I'm on a limited budget and want to make sure I get the most for my money... but I do like to see a checklist of features among different cars which I may be consider buying! ;)

Thanks,

Robert.

Tom,
Excellent response.

First poster. So are you going to buy that car (BMW vs Volvo) based on who lists the most features and divide by the price the dealer gives you? UGH! So if a Volvo has a better parts to cost ratio than a BMW that makes a Volvo a better car? Or if you are a family of 4 and a BMW sportscar (eg M3) has a better parts to cost ratio than a Volvo station wagon or SUV type then an M3 is a better car?

(I don't want to start a car war, neither of these can I afford and I think they are both excellent manufactures.)

>>Maybe I'm strange, but that's the way I shop for a car!

You are strange!
;-)

Hi Tom

PostgreSQL supports these many languages as a procedural extension to SQL
::: PL/pgSQL ::: PL/Lua ::: PL/Perl ::: PL/php ::: PL/Python ::: PL/Ruby ::: PL/sh ::: PL/Tcl ::: PL/Scheme ::: PL/Java ::: PL/R

Oracle provides only PL/SQL (Java Stored Procedures (or) OCI (or) etc. cannot be compared to the PL/SQL, I think. They still require PL/SQL wrappers.)

I think both the approaches of Oracle and PostgreSQL have their own advantage and disadvantage
(·) Many languages may result in confusion. For a PL/pgSQL developer, PL/Python may look very strange. So, hard to maintain the codes. This would result in negative effects for an organization.
(·) Many languages means the open-gate for a wide range of developers. A Perl developer would find PL/Perl much easier. A Scheme developer would find PL/Scheme much easier. This would attract a lot more programmers towards PostgreSQL than Oracle.

May I know your opinion about this?

Thanks In Advance

I have performed some test with PostgreSQL few months ago and have noticed a very fascinating thing - its insert performance is practically unaffected by the indexes on the table - by Oracle it degrades significantly by the index addition. On the table w/o indexes PostgreSQL was much (times) slower, on the table with 3 indexes - faster then Oracle, at least in my tests.

I am wondering how they achieved such behavior.

postgresql docu seems to state the opposite:
http://www.postgresql.org/docs/8.0/static/populate.html#POPULATE-RM-INDEXES

@OP: Actually wikipedia has a basic comparison of various RDBMS systems by quite several aspects.
But as Tom suggested first make your wish list, then look for the best matching candidate. As always with wikipedia don't forget some own research on the last bits of information as not everything is accurate or up to date. But it will give you a good start.

regarding insert performance: Postgresql uses a different MVCC modell which does not cover indexes. This reduces the amount of data that needs to be written before returning a commit. The price you pay is that no query ever can be answered from the index itself: Postgresql always needs to look at the table row for MVCC information.

Oracle implements MVCC at a quite low level using undo tablespace for just everything. And undo is protected by redo. The price is additional IO for every commit. The benefit is that just everything works magically with MVCC as people expect it without thinking about it.
You will suffer from this model using autocommit. But then you might not need an ACID compliant RDBMS at all.

My guess is that the original poster is trying to justify any costs of using Oracle to his boss (who's probably saying, hey, here's something cheap over here, why not use that?).

Not bashing postgreSQL, but I've seen this situation before. Budget concerns leads to judgments based on price alone, and the company later finds out their db doesn't have the features (or doesn't scale, or whatever) like Oracle would (and gee, why can't we find qualified DBAs or developers on this platform?). Eventually, the question becomes, how can we migrate our backend systems over to Oracle?

So, I guess I'm saying that saving money initially may mean losing money over the long run. (and no, I don't work for Oracle ;)

Best of luck.

The free PostgreSQL database seems to be 10 to 20 years behind Oracle. Our Oracle databases perform 20 times faster than our PostgreSQL databases.

Oracle allows you to suggest or force execution plans to be chosen by the optimizer using half a dozen techniques. The PostgreSQL community doesn't even understand why you would want to. They act like optimizers always calculate the cardinality correctly and pick the best execution plan.

You can join 25 tables together in Oracle and use the HASH hint when the cardinality is skewed by multiple filters and join conditions.

In PostgreSQL you have to create a dozen temporary (intermediate) tables to get a reasonable execution plan which still runs 4 times longer than Oracle to get the same result.