[aprssig] High speed datbase inserts <thank you>

[Andrew Rich (VK4TEC)]

Thank you to all 

I have tried insert delayed, will tune my mysql as well !

Cheers 

On Tue, 2005-03-01 at 07:46 -0500, Steve Dimse wrote:
> On 3/1/05 at 4:35 PM Andrew Rich <vk4tec at tech-software.net> sent:
> 
> >I have been experimenting with writting to files every 10 minutes and then
> >doing inserts into mysql
> >from these files. With the idea of giving the database time to breath for
> >searches.
> >
> Unnecessary. If you have an application where the selects take a long time, and
> there is not a need to accesss new data immediately, then you can use INSERT
> DELAYED to intelligently batch the inserts. This is far more efficient that what
> you are proposing.
> 
> http://dev.mysql.com/doc/mysql/en/insert-delayed.html
> 
> >I have some questions regarding perl. In the duration I am closing off one
> >file and opening up another,
> >would the telnet data be "pooled" by Net/Telnet.pm.
> >
> Not sure what you mean by pooled. If you are asking if the data is buffered, yes
> it is, you would not lose data while the program was doing a close/open.
> 
> >Would I be better off inserting straight into the database ? and not worry
> >about files ?
> >
> Yes.
> 
> >How does findu.com do it ?
> >
> directly, and without DELAYED, I want new data to become immediately available.
> On average over the last month, findu's MySQL averaged 106 transactions per
> second, 30% inserts and 60% selects, with a 99.63% key efficiency. You can get
> these stats with a program called mytop, like the unix top, but for MySQL. 
> 
> You must tune MySQL to the insert/select mix of your application, out of the box
> it is set up to be easy on system resources, if you are running it on a box
> largely dedicated to the database, you need to tell MySQL it is OK to grab more
> resources. There are books written about MySQL optimization, a good place to
> start is the MySQL chapter:
> 
> http://dev.mysql.com/doc/mysql/en/mysql-optimization.html
>  
> Steve K4HG
> 
>