Hi,
Mark Callaghan wrote:
> On Wed, Oct 8, 2008 at 8:58 AM, Stewart Smith <stewart@stripped> wrote:
>
>> On Tue, Oct 07, 2008 at 12:51:49AM +0300, Nick Keefen wrote:
>>> hi Jeff.
>>>
>>> I'm impressed. I figured that was the optimizer's fault and tried to
>>> change its depth but it did not help. STRAIGHT_JOIN gives the perfect
>>> performance of split seconds!
>>>
>>> Unfortunately it might be tricky for me to modify my webapp (i'm more of
>>> a system administrator and am uncomfortable with the idea of modifying
>>> the client's code). Is there any way to make the query optimizer work
>>> correctly?
>> Not without modifying the query. MySQL doesn't have a way to save/change
>> query plans for specific queries.
>>
>> The best thing would be to contact the app authors and point this out to
>> them.
>>
>> We have work in progress on getting index statistics to mysql optimiser
>> that should dramatically improve things. Although this isn't a solution
>> to your problem today, it may well be in the future.
>
>
> By this do you mean that a storage engine will be able to provide something
> better than index cardinality stats with the assumption of uniform
> distribution?
>
No, he's referring to NDB making proper use of the existing interface.
However we'll soon start work in this direction towards giving the
engine possibilities to report more refined statistics as well as the
optimiser itself gathering such statistics.
Rgrds Mikael
>
>> --
>> Stewart Smith
>>
>> --
>> MySQL Cluster Mailing List
>> For list archives: http://lists.mysql.com/cluster
>> To unsubscribe:
>> http://lists.mysql.com/cluster?unsub=1
>>
>>
>
>
