Resource Plots

Altair Accelerator tools save information about the utilization of each resource map, which can be displayed in graphical form. Two plots are generated for each resource: one plot shows the utilization of the resource; one plot shows the demand for the resource.

The plot can be generated over a period of 1, 2, 4, or 24 hours. Options are available to zoom in or out of the view, and to pan the view. You can focus on the details of the selected range of time as shown below.

To view a plot, on the browser go to the Resource Statistics page, and then press the graph button that is next to the desired resource name in the Resource column. The Resource Statistics page is shown in Statistical Information about Resources and Jobs.


Figure 1.

Resource Utilization

The first set of plots shows the availability (green background) and utilization (blue line) of a resource in an interval of today > (24 hours). The height of the blue line indicates the usage of the resources: the higher the value of the plot line, the more efficient is the usage of the resources.


Figure 2.

Queued Jobs and Unmet Demand

The second set of plots show the demand for the given resource. In this example, it starts with no jobs in the queue. After a few minutes jobs are started. More jobs are added as previously submitted jobs are executed. about 350 jobs is added. More jobs from the are added as submitted jobs are executed. (In this example, short jobs are submitted.)

Accelerator can relate the jobs in the queue with the utilization of a resource. When a resource is exhausted and more jobs in the queue ask for that same resource, that indicates unmet demand. Unmet demand is represented by the red line in the plot.

When the red line tracks the dark blue line, it indicates the resource is critical: this condition limits the ability of FlowTracer to process the queued jobs faster. When the red line is significantly below the dark blue line, it indicates that the dispatching of the queued jobs is limited. The limitation could be caused by the insufficiency of another resource such as CPUs.


Figure 3.

Use the Plots to Plan for Future Software/Hardware Purchases

If the plotted resource represents an expensive software license, it is important to look carefully at the unmet demand curve.

If the unmet demand is low, presumably there is an excess capacity for this license and the performance of the queueing system would be the same even if this resource was reduced. On the other hand, if the unmet demand is high, purchasing more licenses could be a benefit.

Note: When resource utilization is low, it may indicate there is a bottleneck in the queue and that unmet demand is actually high. For example, there may be a rush-hour of jobs that exceeds the availability of licenses, which then limits the usage of other resources. These situations typically appear as spikes in the graph. The queuing system manages the spikes of license demands. However, for planning purposes, it is important to analyze these spikes and determine if there is a significant effect on project timelines as well as resource utilization.