Deal better with extremely spikey data
CONTEXT: Some counters displayed by SQL Monitor is extremely spiky. SQL Monitor scales its graph such that the any spike fits on the y axis
PROBLEM: If there is one extreme spike, all the other data gets scaled such that it's very close to the x-axis, making any patterns in the data hard to see.
EXAMPLE SOLUTIONS:
• allow to manually zoom into the y axis
• allow extreme spikes to be beyod the drawing area of the graph (still with some indicator of how high the spike is)
• have an option to show the graph on a logarhithmic scale
What do you think about these solutions? Got another idea? Let us know in the comments!
Anonymous
shared this idea
-
Stefan LK
commented
Currently the max and min values under the statistics tab only looks at the data point in active chart. Should it not use the raw min and max, across all available data points, during the time range?
-
Janet R
commented
the graph is not a lot of use without some indication of what values the lines are representing
-
Blaž Dakskobler
commented
Well, an alternative idea to achieve this (i.e. to eliminate spike noise) is to enchance zooming / dragging in Analysis graph with an option to "lock" X-axis and only zoom/drag along Y-axis (the suggestion was merged here, so I'm reposting it).
-
Thanks a lot for your feedbacl Blaz.
Regards,
Priya -
Mark
commented
An option to change the background color would be useful as well. It's hard to discern the lower and upper boundaries when it's transparent.
-
Thanks for your suggestion.
Regards,
Priya -
Blaž Dakskobler
commented
Indeed - in case of extreme spikes you don't see the details you want to, making the graph useless unless you "zoom in" to a shorter time period, thereby often losing overview of the long term baseline.
-
Anonymous
commented
Maybe a log-scale option.
-
Priya Sinha
commented
Thanks a lot for the suggestion John.
-
John B
commented
Auto scale Y-axis appropriately for selected time range's minimum and maximum