Forum Discussion
Changing the X-Axis range on a custom view?
I'm currently getting "Disk Reads/sec" and "Disk Writes/sec" from the windows performance counters and pulling them into OMS.
If I use the Advanced Editor, the timechart I generate looks great as it begins at the start of my data. When I take the same query, and put it into a line chart within a custom view, the data gets all scrunched up as I am a 1 second interval and the line chart has an X-Axis range of 24 hours.
Is there any way to change the X-Axis range on a custom view?
This is my query:
Perf
| where ObjectName == "LogicalDisk"
and CounterName == "Disk Reads/sec"
or CounterName == "Disk Writes/sec"
| where ( InstanceName != "_Total")
| summarize avg(CounterValue) by bin(TimeGenerated, 1s), InstanceName
My guess is you're overloading your chart with too much data. I just tried it out, and increasing the bin size seems to correct it. 1sec is pretty extreme anyway for a chart anyway. 1min or even 5min uses the same data and should display the chart fine.
7 Replies
- Ketan GhelaniFormer Employee
1) When you say "pulling into OMS" - Do you mean the Log Search page in OMS or the View Designer or overview tile.
2) By Advanced Editor do you mean Advanced Analytics portal - correct ?
3) By Custom View -- I am assuming you created a custom view in View Designer, It doesn't support these many data points.
- Brian Wren
Microsoft
My guess is you're overloading your chart with too much data. I just tried it out, and increasing the bin size seems to correct it. 1sec is pretty extreme anyway for a chart anyway. 1min or even 5min uses the same data and should display the chart fine.
- Tim CurwickCopper Contributor
Brian,
Being able to look at resource usage in 1 second bins is not an unreasonable or unusual requirement. 5 minute bins that show a flat 20% usage of disk or memory or CPU can represent a normal load, or it can represent a normal 10% load plus a bad spike that is pegging the resource for 30 seconds straight once every 5 minutes and having a major impact on end users and one's business. Ops engineers need to be able to look at and visualize historical logs non-aggregated.
- Brian Wren
Microsoft
Actually, bin of 1sec gives me a warning in the Advanced Portal. Averaging on 1min should still show spikes and be suitable for analysis. Most of the perf data isn't sampling at 1sec anyway. Keep in mind that you can also configure alert rules on near real time perf data to quickly identify anomolies as well.