Re: Cache Hit Ratio from system views

>> > Yet you still insist that the two ratios can be
>> > compared 'all else being equal'. You haven't told us what, exactly,
>> > that proves, except that you can calculate a ratio consistently.
>>
>> That proves BCHR is meaningful.
>
> How, precisely, does it prove the BCHR meaningful? Let's say, for
> example, your precious BCHR is sitting at a comfortable 97% and,
> miracle of miracles, your database is performing admirably. An hour
> later your BCHR is STILL sitting at 97%, but now you have a line of
> users outside your door complaining the database isn't responding.
> Your BCHR is at a lovely 97% so everything must be fine, no? No, even
> though nothing in the database configuration has changed (your 'all
> things being equal' noise). Tell us what this BCHR is telling you to
> do to fix this issue? Absolutely nothing. Something changed
> (possibly a user running some one-off query bringing the database to
> its knees, possibly something else) but the BCHR can't show you WHAT
> changed. In fact it didn't even BUDGE from its original value of
> 97%. It was flashing 'green' an hour ago, and it's flashing 'green'
> now, but something is amiss and it can't tell you what.
>

Again you have not read my comments. Someone brought up exactly the same point earlier.

>> It is ironic we are debating about something
>> many consider meaningless.
>
> Apparently all except you consider it meaningless.
>

That's right.

>> Do you ever adjust the buffer cache size?
>
> Rarely, and only when it's actually necessary.
>

So you do adjust cache size.

>> What do you use as reference?
>
> Wait statistics, usually. The BCHR doesn't enter into that discussion
> for what I consider good reasons which have already been listed in
> this thread.
>

Wait statistics? I am getting very curious. Which wait stat do you use for tuning buffer cache? Received on Thu Aug 30 2007 - 22:38:27 CDT