PMP database Schema

[Robert Gadbois]

I noticed that PMP writes to a different database (I'm sure that there's a good reason for not keeping it simple and more tightly integrated with CIM ...)
On that note, does any one have the schema for the PPM database so that we can run custom reports against it.
Also I noticed an earler thread on not being able to reset the default thresholds on the PMP, has a feature been added to allow this yet?
Robert

[Scott Shaffer]

You are not able to modify the thresholds that PMP uses. They are based on our experiences and are the key part of the analysis. What threshold where you thinking of modifying?

Also we aren't documenting any of the PMP database schema at this time. What would you like to be able to report on?

[Robert Gadbois]

Hi,

I'd like to calculate the peak, average medium processor utilization as well as similiar values for memory and cache utilization; an average figure just isn't meaningfull enough for us.

As for the PMP thresholds their are times when some servers peak which is normal, but I'd like the flexibility of setting a duration of peak for these servers which is normal.

On that note, I'm looking for a best practices document from HP on what are recomended thresholds an traps to set/monitor in a normal environment.
We use CIM, PMP and HP ITO to monitor various areas of our NT/W2K environment but don't want to "over monitor" every server.
Assuming I break my servers into criticality catagories (red, yellow, green), do you have a list of items/traps that you would suggest to monitor?

[David Claypool]

Robert:

An addition to what Scott said and a suggestion...

As you are using PMP already, you can see that it reports and graphs min, max, average and last for the important data points in each subsystem. PMP's analysis (what turns the server orange, red or green) relies heavily not on absolute values like processor utilization (this varies by application greatly) but on how many things are getting backed up in a queue. That is indicative of a bottleneck and if a lot of requests are waiting to be serviced, something is wrong regardless of utilization.

The suggestion is that while IM7 may not gather the data you want, there is a lot of it available from the agents if you want to get it. One popular way is the open source "Multi-Router Traffic Grapher" at www.mrtg.orgthat is available for Windows and Linus. There is a document called "Using ProLiant management agents with open source monitoring tools" at http://h18013.www1.hp.com/products/servers/linux/documentation.html that describes the usage of this tool. It's pretty slick.