see my other reply but note he is waiting on reads not writes (which doesn't
make the system great of course).
--
Niall Litchfield
Oracle DBA
Audit Commission UK
"Jim Kennedy" <kennedy-down_with_spammers_at_attbi.com> wrote in message
news:GTx_9.112464$AV4.3688_at_sccrnsc01...
> Yes, function based indexes are great. You MUST use the CBO, so if this
is
> a vender appliction (eg Siebel) that requires you to use the RBO then you
> are screwed with that method. In which case you have to resort to the
> phantom column approach. :-( Raid 5 is notorious for a write penalty.
Let
> me guess IT built the box for you with out input from the DBA.
> Jim
>
> --
> Replace part of the email address: kennedy-down_with_spammers_at_attbi.com
> with family. Remove the negative part, keep the minus sign. You can
figure
> it out.
> "Gary Avery" <Gary.P.Avery_at_Pharmacia.com> wrote in message
> news:3e3aa1ff$0$209$1406d58a_at_newsread.pharmacia.com...
> > Dear board,
> >
> > here is some of the output from a statspack report I ran (went on the
> > Performace/Tuning course last week)
> >
> > Snap Length
> > (Minutes)
> > -----------
> > 168.90
> >
> >
> > Top 5 Wait Events
> > ~~~~~~~~~~~~~~~~~ Wait %
> > Total
> > Event Waits Time (cs)
Wt
> > Time
>
> -------------------------------------------- ------------ ------------ ---
> --
> > --
> > db file sequential read 524,735 235,947
> > 76.07
> > db file scattered read 157,554 48,837
> > 15.75
> > db file parallel write 1,993 8,248
> > 2.66
> > db file parallel read 2,217 5,562
> > 1.79
> > control file sequential read 14,152 3,896
> > 1.26
> >
> > If I read this right, then of the 168 minutes that the snap was taken
the
> > highest 2 waits accounted for 47 minutes of the snap. Now to me that
seems
> > an awful lot and I beleive I ought to be looking at what sql is being
> > performed against the tables to see if there are some obvious indexes
> > missing. Is this what i should be doing or are there other 'quicker'
gains
> > to be made elsewhere. (I have already up'd the db_file_multiblock_read
> > count)
> >
> > We run a multi-purpose application with all the datafiles stored on a
> single
> > raid5 array (NT).
> >
> > Also has anyone done much with function based indexes? Would an index on
> > UPPER(surname) gave a dramatic increase to selects...where surname like
> > (upper(blah)). we tend to do a fair amount of those.
> >
> > regards
> >
> > Gary
> >
> >
>
>
Received on Fri Jan 31 2003 - 13:07:50 CST
