Presently reports and graphs are sorted by the most recent polling sample. I would like to request that reports and graphs are sorted based on the average over the entire time period being reported/graphed.
The problem with the present sorting method causes otherwise top ranking devices/hosts to be missed in any Top-N reports. For example a lun that gets consistently high IO utilization through the week but dips down to almost no utilization at say 5 AM after a backup job completes. Lets say a weekly report is generated during this low point of utilization. Even though the lun was possibly very busy ( or busiest on an entire array ) 23 hours a day, it might get ranked as the 5th or 50th lun by utilization if it just happens to be idle when the report is run.
The work around offered by support is to generate a report showing only the average over a given period to get it to sort on that period properly. Meaning, if you want a weekly report to sort properly, you must generate a "Last 7 days" report, with Data set to "weekly". This results in a properly sorted "graph" with only a single point of data for the week per device, and no quarter hourly/hourly or daily type of resolution to the graph.
Thank you for considering this feature request.
Please reference support case number 559781.