Analyzing metric data - SQL Monitor

AnalysisTab

The Analysis page charts performance data on a wide range of key metrics for monitored servers and databases, including metrics created by you (see Creating custom metrics and alerts). You can display graphs containing data points that represent points in time when performance data was collected.

To view an analysis graph

Choose the performance metric you want to plot from the Show drop-down list.
Metrics are grouped into Machine, SQL Server, Database and Custom metric categories.
Select the object to plot. For some types of metric, you can plot all the objects together. The example below shows Machine: processor time for all machines in the clusters on the selected server.
Choose a time range. By default, the graph shows the last 10 minutes, but you can select a different fixed value from the Time range drop-down list.
To specify a particular time range, click the Set custom time range icon and enter the start and end time.

Comparing time ranges

You can also use the Time range drop-down list to compare how your servers were performing in relation to the same time period from the last hour, day, week or month. Your options for comparison are as follows:

Analysis_page_time_range

Viewing values on the graph

Each graph line represents the performance of a selected object. SQL Monitor plots up to a maximum of 500 data points on each graph and draws straight lines between consecutive points. For short time ranges (less than 10 minutes) the graph displays fewer data points. For longer time ranges, more data points are displayed, so that these graphs can still provide a detailed overview of changes in values. Move your mouse pointer over a line on the graph to view a tooltip showing the time, object and value at that data point.

If you are comparing data for different time ranges, the graph line for the older time range is displayed as a dashed line. If you are comparing data for multiple objects, a color-coded key is displayed below the graph.

AnalysisTooltip

Viewing statistics

Click the Statistics tab to the right of the graph to display the mean, maximum and minimum performance metric values for the selected time range. If you're comparing data at different time ranges, the difference between each value is also displayed.

Exporting the data

Click Export... to open or save metric data as a .csv file for the object and time range specified. This means you can record object data for use in reports on performance. The export.csv file contains:

the name of the selected metric and host machine.
the type of value displayed if this doesn't form part of the metric name, for example, bytes, milliseconds or percentages.
a column containing the date and time of every data point displayed on the analysis graph. The time is displayed as Coordinated Universal Time (UTC).
columns containing details of every data point for each selected object.

Note: If necessary, you may consider adjusting the UTC times displayed in the file according to the time zone in which your servers are located.

Finding more information about metrics

Click the Description tab to the right of the graph to display detailed information about the selected metric, including:

the name of the equivalent PerfMon counter
an explanation of what this metric measures
guidelines for acceptable values, and values that may suggest a problem with your object
related metrics that should also be checked
suggestions for solutions to problems caused by unacceptable values
links to articles or white papers containing additional information

Plotting data points from collected values or calculated values

Data points are plotted differently depending on whether values are collected or calculated:

collected values - data points represent the actual collected values. For example, Disk used simply displays the raw values provided by PerfMon. To optimize performance, some data points are not plotted if a long time range is selected.
calculated values - SQL Monitor has to carry out its own calculation on raw data to generate a value. For example, to work out Transactions/sec, the number of transactions between data points is divided by the number of seconds between data points. This averaging has the effect of smoothing out fluctuations.

Data missing from the graph?

SQL Monitor needs at least one data point to plot the graph. Data points may be missing for the time selected for the following reasons:

data for that time range has been purged.
the monitored object was unavailable or suspended.
SQL Monitor was not running at that time.
you are analyzing a custom metric measuring a rate of change, but not enough data has been collected to calculate the rate. The rate is calculated from two consecutive collection values. The first collection takes place immediately after the custom metric is created. The second collection takes place according to the defined metric collection frequency. Check the graph again when at least two collections have been made.

If data points cannot be plotted on the graph, no line is displayed between the last two successfully plotted points.

Was this article helpful?

Search support