[askdba] Re: Buffer cache hit very low...!!!

  • From: Chirag DBA <chiragdba@xxxxxxxxx>
  • To: davidsharples@xxxxxxxxx
  • Date: Tue, 13 Sep 2005 17:19:43 -0400

This is the topic I have seen long long discussions but never ended up with 
a single voice.
 What I think is the main reason for performance is the bad SQLs and the 
next is index is not working properly. Might be because of getting split. 
Here we have lots of DMLs so might indexes are got split and are leading to 
FTS of the tables.
 and I have seen these things really working and that is why I think I still 
believe this strongly.

 On 9/13/05, David Sharples <davidsharples@xxxxxxxxx> wrote: 
> well you rebuild your indexes (why are you doing this - saying FTS doesnt 
> make sense as FTS's dont use indexes)
>  if you rebuild indexes it will flood you buffer cache with rubbish so 
> thats why your ratio is low, so you spend all morning doing PIO's which slow 
> your users down.
>  You seem fixed on rebuilding your indexes for everything, you may want to 
> review that policy
>  On 9/13/05, Chirag DBA <chiragdba@xxxxxxxxx> wrote: 
> > 
> > Still,
> >  Well, Users are happy. but in the morning when they run large batch 
> > jobs, yeah couple of guys do face some problems. As per my observations 
> > ,highest waits were DB_SEQUENTIAL_WAITS.
> >  So have done rebuilding of indexes as Sequential reads can be high 
> > because of the FTS on the tables. Even for the 40GB Database, we have only 
> > 250 MB SGA so which I am planning to increase. In normal day operation no 
> > query is getting locked or taking very long to execute. 
> >  

Other related posts: