|
Trend Graph
Graphs showing the measurements over time are ideal for real-time analysis of metrics. For capacity planning or for long term historical analysis, trend graphs that denote trends in the data values over time are more appropriate. The eG Enterprise manager computes trend values from the raw measurement data as part of its day-end activity. Trend values are computed on an hourly, daily, and monthly basis. Since these values represent the results of the analysis of the raw measurement results, the trend values take up much less storage space in the eG database than the measurement values. The trend graph option in the eG monitoring console allows users to view and analyze the computed trend values over time. Besides revealing normal performance levels, trend analysis also aids administrators in forecasting future usage trends and accordingly planning the capacity of target servers.
To access the TREND GRAPH page, select the Trend option from the Graphs menu. The
following steps have to be done to plot trend graphs
-
The
service for which the graph is to be plotted can be selected from the Service
list box. The Component
list box gets populated with all the components associated with this
service. However, this step is optional. If the user does not select a
service, all the components in the infrastructure that are currently being
managed by the eG manager will be listed in the Component
list box.
-
Next,
a specific component associated with this service can be selected from the Component
list box. You can even narrow your component-search further by specifying the whole/part of a component name in the Search text box here and clicking the right-arrow button next to it. All components with names that embed the specified search string will then populate the Component list, thus making it easier for you to pick a component.
-
If the chosen test is being executed by more than one measurement host, then the Msmt Host list gets populated with all such measurement hosts. From this list, select the measurement host for which the graph is to be generated.
-
Next, indicate the type of trend graph to be generated - whether 3D or 2D - by choosing the appropriate option from the View list. The default selection in this list is determined by the Show monitor graphs in 3D flag in the MONITOR SETTINGS page of the eG administrative interface. If this flag is set to Yes, then the 3D option will be selected by default in the View list. If this flag is set to No, then the default selection in the View list will be 2D.
-
The
period for which the variations of the selected measurements have to be analyzed can be specified using the Timeline list. You can either provide a fixed time line such as 1 hour, 2 days, etc., or select the Any option from the list to provide a From and To date/time for graph generation.
-
By default, the eG manager uses statistical quality control techniques to compute the upper and lower bounds of each metric in the database. This computation removes the outliers among the data values for the chosen period. The default Min/Max selection from the Graph list shows the upper/lower bounds for each measurement over time. In some cases, capacity planners may prefer to view hourly/daily/monthly averages of metrics. The Average option in the Graph list displays such a graph. For instance, you can plot a trend graph that depicts how many user connections on an average were established with a critical Terminal server, every day during a couple of weeks; besides indicating the normal load on the Terminal server, such a graph also enables you to understand whether the Terminal server has been adequately tuned to handle higher loads, and thereby helps you make effective sizing recommendations for the future. Many of the measurements made by eG agents represent data values over time - eg., the rate of requests handled by a web site, the rate of traffic to and from a network node, etc. While data rates are good indicators of the workload from a monitoring perspective, from a capacity planning perspective a better indicator would be the total volume of requests or data handled. In the case of a web site, for example, the total number of requests handled by the web site per hour, day, and month would be of interest. The Sum option of the trend graph provides this capability. If the unit of the selected measure is a number - eg., requests handled in the last measurement period - the Sum graph plots the sum of the measure values over time. If the measure is a rate (e.g., Request/second handled by the web site), then eG Enterprise manager converts the rate into an absolute value by multiplying the rate with the measurement period, and computes the hourly, daily, and monthly sums of these absolute values during the trending process. The results of this computation can be plotted over time in the Sum graph.
Note:
The capability of the eG manager to compute the Sum and Average of metrics is governed by the Compute average/sum of metrics while trending flag in MANAGER SETTINGS page (Configure -> Manager Settings) in the eG administrative interface. By default, this flag is set to Yes indicating that eG Enterprise computes and stores the average and sum values of every performance metric in the database, by default. If, for some reason, you want to disable this capability, just set this flag to No, and Update the changes.
Unlike average values, Sum computation could be meaningless in the case of a few measures (eg., Availability, Percentage usage of disk space, etc.). Therefore, instead of maintaining the Sum values for all measures in the database, you can configure the eG manager to compute Sum for specific measures only. To do so, follow the steps given below:
|