DCE Trending Points Calculation/Issue

[DCIM_Support]

I have a concern with the way DCE displays the Trends of a device and want to know how the points shown are chosen. How are the trends points which are shown calculated with respect to the x axis time length?

If I look back 24 hours I see a temp spike, which I should because there was one. There was also one three months ago because of an issue. When I just view the 24 hour trend I see the spike however when I change the range to 4 months I do not see either spike.

Since I know when the spikes occurred from other logs I can go back into the trend from three months ago and just look at the 24 hour period around that spike and see the spike in great detail. So the data is there.

So how are the trends points which are shown calculated with respect to the x axis time length? To me, when the trend is viewed over a long period of time the MAX and MIN points should be shown first vice just showing a point every hour. I can then see when my issue points are and focus on those.

Regards,

J

(CID:110008139)

[DCIM_Support · Accepted solution]

Hi John,

I can't provide the exact algorithm however k-base FA158422 does I believe address the concerns you are having. If the graph is not including the peaks and/or valleys you're concerned with, I believe the only option is to graph fewer sensors or a smaller time period.

Thanks,

Steve.

(CID:110008166)

[DCIM_Support]

Steven, thank you. That does explain my issue with showing only 10,000 data points however I think this is flawed when it states : "Minimum or maximum data points for isolated spikes or drops may be eliminated." This is sort of what I am looking for. And it states "isolated" which is good. but there are whole events (spikes) which grow over an hour or so which are missing. I want to know where there were events (spikes or drops). I don't really care that the values stay the same. I need to know where there are issues. I am not sure how to fix it. However I can tell you when I look at trends over months and see no spikes or dips, it is misleading.

(CID:110008185)

[DCIM_Support]

I want to bring this back up as it is really annoying/un-useful. I use the 30 day graph to look for loading on a PDU and I miss the spikes in load which cause shutdowns...

The trends are basically useless in this format...

I think others would have the same concern if they knew how the graphing works and how it misses the peaks and valleys which really matter...

(CID:134687518)

[DCIM_Support]

This question is closed for comments. You're welcome to start a new topic if you have further comments on this issue.