This post is more than 5 years old

1 Rookie

 • 

66 Posts

1445

June 11th, 2015 06:00

Utilization Projection - Data Domain

How long must one wait, how much data must be in the system, to be able to predict data domain utilization over time?

I've tried building my own reports, and running stock reports. I can't get any that will average the utilization over the last week, and project that average change into the following week.

I have about 2 weeks worth of data in the system at this point.

DPA 6.2.1

Data Domain 5.4

Networker 8.0.4

And yes, I downloaded and eventually figured out (maybe) how to use the damned data domain import tool. (Which is really 3 scripts with no information on how they have to be used).

I've run it twice, and it doesn't seem to have any impact on the report I'm looking to run.

141 Posts

June 11th, 2015 06:00

Yeah try changing the minimum trend fitness down to 0.1 (from 0.8) and see if it helps.

Yes to predict 3 months ahead, it is recommended to have 6 months worth of data. It will probably work without that much history but that is the recommendation to try and make the prediction as accurate as possible.

Thanks

David

141 Posts

June 11th, 2015 06:00

Hi,

In order to trend into the future it is recommended to go back twice as far as you want to go forward. So with 2 weeks of data, you should be able to predict 1 week ahead. By default the settings are set that you need 10 data points with a minimum fit of 80%. You can adjust these settings but it will affect the accuracy of the trend. These 2 settings can be found under your User Properties, under the Report Preferences tab and are called "Minimum Trend Points" and "Minimum Trend Fitness".

You shouldn't need to use the Data Domain Processor Tool to achieve this report, as the data is gathered by the standard requests.

HTH

David

1 Rookie

 • 

66 Posts

June 11th, 2015 06:00

Interesting... perhaps the problem then is the minimum fit, I'm trying to predict data domain utilization It hovers between 55 and 65% the weekly cleaning drops it by as much as 10%.

To predict 3 months forward, I would probably need 6 months of history?

No Events found!

Top