Subject | AW: Re: [Firebird-Java] How to change forced writes settings via JayBird? |
---|---|
Author | Roman Rokytskyy |
Post date | 2009-10-13T08:11:58Z |
If I remember correctly, firebird.conf is read from the directory specified in %FIREBIRD%, so you definitely can overwrite the settings for the test environment.
Roman
----- Ursprüngliche Nachricht -----
Von: Alec Swan <alecswan@...>
Gesendet: Dienstag, 13. Oktober 2009 06:24
An: Firebird-Java@yahoogroups.com
Betreff: Re: [Firebird-Java] How to change forced writes settings via JayBird?
Is there any way to override these settings in my test environment beside
changing them in firebird.properties? Are there any API I can call or system
or environment variables I can set to accomplish that?
Thanks,
Alec
Roman
----- Ursprüngliche Nachricht -----
Von: Alec Swan <alecswan@...>
Gesendet: Dienstag, 13. Oktober 2009 06:24
An: Firebird-Java@yahoogroups.com
Betreff: Re: [Firebird-Java] How to change forced writes settings via JayBird?
Is there any way to override these settings in my test environment beside
changing them in firebird.properties? Are there any API I can call or system
or environment variables I can set to accomplish that?
Thanks,
Alec
On Mon, Oct 12, 2009 at 7:22 PM, Helen Borrie <helebor@...> wrote:
>
>
> At 11:35 AM 13/10/2009, you wrote:
> >Can anybody explain the reason why ForcedWrites is a database property and
> >MaxUnflushedWrites is a server property given that the later is only used
> >when the former is set?
>
> MaxUnflushedWrites is a parameter that dictates the server's interaction
> with the operating system, not the server's interaction with databases. It
> is only applicable to databases that have ForcedWrites off, which is about
> the server's interaction with that particular database. It's more "usual"
> than "unusual" to have FW on for all databases. It's normally a temporary
> usage for a database that's about to undergo a huge batch operation, since
> it can allow such operations to go faster.
>
> FW is not something to disable lightly. On Windows particularly,
> asynchronous write adds a high level of risk of corruption that is otherwise
> absent. FW is of no help whatsoever in improving performance that is slow
> due to inefficient metadata structures or indexing, inefficient queries,
> heavy garbage buildup, inappropriate cache size, inadequate system
> resources, overlooked index maintenance, etc. etc. - hence Roman's comment
> that you interpreted as "sarcastic".
>
> Helen
>
>
>
[Non-text portions of this message have been removed]
------------------------------------
Yahoo! Groups Links