Pantek Library
Hosting Provided By
CybrHost
High Speed Hosting

Re: [patch 1/1] WL4092 Remove ndb_use_exact_count - use cache of approximate count instead

From: Jonas Oreland <jonas(at)mysql.com>
Date: Fri Oct 12 2007 - 02:27:50 EDT


Stewart Smith wrote:
> On Fri, 2007-10-12 at 07:40 +0200, Jonas Oreland wrote:

>> I propose the following (which should not be a huge extra effort)
>>
>> 1) add a ndb_statistics_check_time (as a mysqld and session variable)

>
>
> An interesting thing with this is:
> - we get row count for free on the get_table_statistics call that the
> util thread does. So ndb_statistics_check_time would only be relevant
> for when: a) query cache is disabled or b) is < ndb_cache_check_time.
> - This could be confusing. Thoughts?

maybe yes, if we *only* have row-count as statistics... but suppose we add more...

conclusion: i think that cache_check_time and statistics_check_time are 2 conceptually different things. and should be treated so...even if they with the current "poor" statistics intersects

>> 2) create a "struct ndb_statistics_cache" (or similar)
>>    instead of share->approx_records), so it's easy to add more stuff

>
> could be good...
>
>> 3) make "analyze table A" force an update of the statistics for A

>
> Seeing as the updates are fairly regular, is this really needed?

yes, if I set ndb_statistics_check_time = 100years and ndb_cache_check_time=200years

>

>> 4) have pekka do his index statistics using same machinery

>
> Pekka, does this fit in with your ideas?
>
-- 
MySQL Code Commits Mailing List
For list archives: 
http://lists.mysql.com/commits
To unsubscribe:    
http://lists.mysql.com/commits?unsub=lists@pantek.com
Received on Fri Oct 12 02:28:34 2007

This archive was generated by hypermail 2.1.8 : Thu Jul 03 2008 - 09:47:59 EDT


Contact Us  Legal Notices  Order Services Online 
Pantek Home  Privacy Policy  IT news  Site Map  Pantek Library