List:General Discussion« Previous MessageNext Message »
From:Jeremy Zawodny Date:June 30 2004 12:53am
Subject:Re: Memory to Memory INSERTS
View as plain text  
On Tue, Jun 29, 2004 at 08:46:35PM -0400, Alejandro Heyworth wrote:
> Eric,
> 
> I'm looking for a way to eliminate the construction, transmission, and 
> parsing of the long multi-row INSERT queries that we are issuing from our 
> client app.  Since we are inserting 200k rows a shot, we're looking for 
> every boost that we can find.
>     * Connecting: (3) [want to use a connection pool]
>     * Sending query to server: (2) [want to eliminate]
>     * Parsing query: (2) [want to eliminate]
>     * Inserting record: (1 x size of record) [no way around this]
>     * Inserting indexes: (1 x number of indexes) [no way around this]
>     * Closing: (1) [want to use a connection pool]
> Since we have already tuned the server, I'm looking for other ideas.
> 
> Radical ideas are welcome!

I missed the earlier part of the thread, but have you considered
simply building raw MyISAM data files (.MYD) from your application?
You could use a merge table over top of them after using myisamchk or
ALTER TABLE to add the appropriate index(es).

Just a thought.  The file format is documented and not terribly
difficult for some applications.

Jeremy
-- 
Jeremy D. Zawodny     |  Perl, Web, MySQL, Linux Magazine, Yahoo!
<Jeremy@stripped>  |  http://jeremy.zawodny.com/

[book] High Performance MySQL -- http://highperformancemysql.com/
Thread
Memory to Memory INSERTSAlejandro Heyworth25 Jun
Re: Memory to Memory INSERTSAlejandro Heyworth29 Jun
Re: Memory to Memory INSERTSSGreen29 Jun
  • Re: Memory to Memory INSERTSEric Bergen30 Jun
    • Re: Memory to Memory INSERTSAlejandro Heyworth30 Jun
      • Re: Memory to Memory INSERTSJeremy Zawodny30 Jun