Comparing Speeds of 3 Databases

Hi, Alexander Predl has made a comparison table of database speeds. This was a quick job in order to give us the direction to go for a new project. Surprise, surprise, HFSQL is faster than PostgreSQL and MySQL



Kind regards,
Guenter Predl
office@windev.at

Hi

Interesting - but I have absolutely no idea what you are testing nor that the columns mean!

Schreiben I assume means write. And Lesen read?

What do the numbers mean?

If it is writes per second it looks like MySQL wins.

Care to explain?

Cheers
André

Hi Guenter,
we always knew that HFSQL is fast
BUT
If you have a lot of users or complex Queries then I am afraid that you will see different results.

Regards
Steven Sitas

comparing speed between DB is a waste of time, IMHO...

the reason is that tests will ALWAYS represent a very small subset of everything that happen in a DB in real life...

Say that a select query is faster in DB A than in DB B...
Faster when run once with the DB doing nothing else
What will happen when you have 300 users doing "stuff" (ie random reading, writing, queries, etc)
What will happen when you have 300 users doing specific stuff (all of them writing at the same time)
what will ...

There is no way to test all those cases, which means that the only significant test can be done IF YOU HAVE A SPECIFIC problem...

I remember talking to a UK user about his application doing security surveillance with thousands of oieces of hardware sending data all the time. He needed to write huge amount of data in real time continuously and he tried to do that with all the DB on the market...

ALL of them were faster than HFSQL at the beginning, and ALL OF THEM crashed after some time, when their fancy buffers were full, while HFSQL continued to write at the same speed all the time...

So for him, in this specific case, HFSQL was the answer.

For me, this kind of test is the only one that has any value.

Generic tests are an answer to a question nobody needs answered.

Hi, it's a set of operations which has been tested, 1000x reading (= Lesen) records, 1000x Writing (= Schreiben) a record and 1000x deleting (= Löschen) a record. For H-commands and SQL commands for three different databases (HFSQL, Postgresql and MySQL) and in a local (on same computer = "Lokal") and a remote (= Extern") setup. Of course, it says something but not everything.

For me, one thing is sure: H-commands are faster than using SQL - for all of the databases. However, remember the program used for the test is a WINDEV .exe. Using C++, Java or Dart will - most probably - gain different results, at least on the SQL side.

There are many more cases to be tested but this one is the first test I've seen to compare these three databases for a WINDEV application. You're invited to write other tests ...

Kind regards,
Guenter Predl
office@windev.at