Re: Metrics for Management

Vince Laurent wrote:

> I got this e-mail the other day from one of our system support folks:
>
> "MANAGER asked me to identify Oracle metrics collected by OVO to be
> displayed on the Metrics database web page. I'm hoping that the three
> of you could give some direction on which metrics may by useful. If
> you have time, scan the list below and let me know which of the
> metrics could be used for capacity planning or fault prevention.
>
> Here are some I selected (let me know if you agree or disagree with
> their usefulness)
>
> # of segments approaching max extent (E017_SegMaxExtentCnt)
> Disk sort rate (E019_SortDiskRate)
> Total buffer cache hit % (E022_TotBufCacHitPct)
> Rate of background checkpoints completed (E035_BckgndCkptRate)
> Table spaces with low free space ; drill down (E206_TblSpFreePct)"
>
> What I am trying to figure out is what are good metrics both to keep
> MANAGEMENT happy as well as help me do my job? Right now I track
> database growth and number of users. What are any of you doing out
> in the real world?
>
> Thanks!
> Vince

1. Number of days since the same day last year divided by the number of cards in a deck minus the Jokers.
2. Amount of time wasted fixing problems created by poor QC practices
3. Number of ambiguous requests for metrics without a clear definition of what information is meaningful to those requesting it

The request is too ambiguous. What I would consider valuable as metrics would be unintelligible to most managers. Send back a message asking that the metrics for the report be specified as follows:

1. What metric
2. Over what time period
3. With comparison to other measurements over what time period
4. With what level of granularity

If the answer is that they don't know ... then go with my first three suggestions. And make the green slice really big and the red slice really small.

Daniel Morgan Received on Thu Jul 18 2002 - 10:25:20 CDT