try to use the same ssd and install windows server 2012r2, windows
server 2016 or windows server 2019 and try it again and you will see
what i talk about
Mit freundlichen Grüßen / Kind regards
Holger Klemt
Am 03.03.2020 um 14:10 schrieb Carlos H. Cantu:
Re: [Firebird-devel] Interbase vs Firebird by Embarcadero Hi Holger,
I got curious on this: "Using nvme ssds on windows server OS without
proper driver support always results in very poor performance"
I have Samsung EVO nvme ssd in my Windows 10 development machine and
never noticed any bad performance accessing FB databases.
[]s
Carlos
http://www.firebirdnews.org
FireBase - http://www.FireBase.com.br
Sean,
on the machine i tested it and in the config i tested based on what i
was able to find from their description
to have a similar test environment, yes, the resulting number that was
reported in the test was more or less
the same for both platforms, but i expect, that this was more or less
a simple result of the maximum I/O
the machine was able to deliver and not really a real world useable
database server test, since caches need
to be filled before the real test starts etc. Our benchmark does a
much better real world result.
i do not remember real result and i already deleted the setup but in
fact on our machine it was even with
both platforms alsmost twice as high as the ones embarcadero showed in
their interbase result and i had
no idea, what they were using to get the bad result for firebird. But
one simple way is know: Using nvme ssds
on windows server OS without proper driver support always results in
very poor performance, perhaps the
combination with journaling makes this advantage, since write I/O
profile is for sure completly different,
but since we know how to make also nvme ssds in windows server os
really fast, the test result is still b...s...
if you and others are interested in my setup, i will have a look if i
can find the tpc setup again on one of my
dozens of tb on several nas systems, just let me know and i can
prepare it again and firebird project can use
that for giving the answer to the embarcadero propaganda.
And in fact, the discussion about this was in August 2017 alread, that
was the first time embarcadero was
announcing "their" tpc "results".
my resume in that discussion then was
"with forced writes on interbase was 0.5% faster, with forced writes
off interbase was 2% slower. "
But as visible in network protocol this was tested with fb3 default
otw encryption on which is not supported by interbase afaik
and when looking at the network packages with a sniffer, it was
obvious that interbase protocol was not encrypted.
I did not retest it with other optimisations in firebird since i knew
already published results from embarcadero
were manipulated or simply not honest.
and for everyone who is interested: feel free to use our benchmark db
and its procedures with interbase database
encryption activated , but don´t expect the result to come fast ...
With a Firebird encryption module active (like the one we deliver in
ibexpert full version), you see a big decrase
in performance when using very low cache buffer settings, but it is
getting almost to full speed if large cache buffers
are used, since the firebird way is only affected on reading and
writing pages, which is obviously not done
so often when data can stay in ram. In Interbase they definitly use
another way, since our results showed
also with large cache buffers the database operations remain much
slower ...
and for Karol,
Delphi is definitly not a bad product, but please understand what my
biggest criticism there is: Each new version
is not only update and work on, there are so many (perhaps intended)
incompatibilities with each new version,
and things like late or missing 64bit android support for native
platforms and other problems in the last years
with their mobile platform support and strange level of half linux
support, and when they talk about multiplatform
but are not even able to have a 64bit Windows IDE, that is not what
the customers want.
We stopped using delphi for all new projects about 6 years ago and
switched to lazarus (with pas2js/tms webcore
for mobile platforms) and we will never go back again, but that
decision is not so easy for existing projects (as
you might know, ibexpert has about 1.7 millions lines of pascal code
and we still use delphi5 to compile and
develop it, even that about 75% of sourcecode is already compatible to
lazarus, the remaing 25% is not
so easy converted in another 25% of time, but we are working on this ...
But in a lot of consulting jobs, we are still working in customer
environments on newer delphi versions so i
know what we are talking about. If anyone is happy with delphi, i
have no problem with this, but i am
extremly happy with my in several thousands of hours used lazarus
development environment which
i was able to learn in almost 0 time because pascal/lcl/vcl is almost
exactly the same in both platforms
and i am too old and too lazy to learn new programming language. And
the advantages of mobile apps
based on lazarus/pas2js/tms webcore is extremly helpful for what we need.
Mit freundlichen Grüßen / Kind regards
Holger Klemt
Am 02.03.2020 um 21:57 schrieb Leyne, Sean:
Holger,
Please clarify
i had discusses their benchmark results already with a embarcadero
?employee in a german forum and they insisted that tpc was bad
?with firebird and good with interbase.
?
?When i reproduced their description of the test on a known powerful
firebird
?machine and exactly the same hardware and same OS with Interbase, the
?results were not really different.
?
?Do you mean that Interbase and Firebird performance was equal?
?
?
?Sean
?
?
?Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
?
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
