5.7

Main Concepts of vCenter Operations Manager
vCenter Operations Manager uses certain concepts that can help you understand the product, its interface,
and how to use it.
Attributes and Metrics
vCenter Operations Manager collects several kinds of data for each inventory object. For example, for a virtual
machine, vCenter Operations Manager might receive data about free disk space, CPU load, and available
memory. Each type of data that vCenter Operations Manager collects is called an attribute. An instance of an
attribute for a specific inventory object is called a metric. For example, free memory for a specific virtual
machine is a metric.
For each metric, vCenter Operations Manager collects and stores multiple readings over time. For example,
the vCenter Operations Manager server polls for information about the CPU load for each virtual machine
once every five minutes. Each piece of data that vCenter Operations Manager collects is called a metric value.
Dynamic Thresholds
vCenter Operations Manager defines dynamic thresholds for every metric based on the current and historical
values of the metric. The normal range of values for a metric can differ on different days at different times
because of regular cycles of use and behavior. vCenter Operations Manager tracks these normal value cycles
and sets the dynamic thresholds accordingly. High metric values that are normal at one time might indicate
potential problems at other times. For example, high CPU use on Friday afternoons, when weekly reports are
generated, is normal. The same value on Sunday morning, when nobody is at the office, might indicate a
problem.
vCenter Operations Manager continuously adjusts the dynamic thresholds. The new incoming data allows
vCenter Operations Manager to better define what value is normal for a metric. The dynamic thresholds add
context to metrics that allows vCenter Operations Manager to distinguish between normal and abnormal
behavior.
Dynamic thresholds eliminate the need for the manual effort required to configure hard thresholds for
hundreds or thousands of metrics. More importantly, they are more accurate than hard thresholds. Dynamic
thresholds allow vCenter Operations Manager to detect deviations based on the actual normal behavior of an
object and not on an arbitrary set of limits.
The analytics algorithms take seven days to calculate the initial values for dynamic thresholds. Dynamic
thresholds appear as line segments under the bar graphs for use metrics on the Details page and on the
Scoreboard page. The length and the position of the dynamic threshold line segment depends on the calculated
normal values for the selected use metrics. Dynamic thresholds also appear as shaded gray areas of the use
metrics graphs on the All Metrics page.
Hard Thresholds
Unlike dynamic thresholds, hard thresholds are fixed values that you enter to define what is normal behaviour
for an object. These arbitrary values do not change over time unless you change them manually. You cannot
fix hard thresholds with vCenter Operations Manager.
Key Performance Indicators
vCenter Operations Manager defines attributes that are critical to the performance of an object as key
performance indicators (KPI). KPI are weighted more heavily in the calculations that determine the health of
an object. Graphs of KPI performance appear before other metrics in several areas of the product.
VMware vCenter Operations Manager Getting Started Guide
8 VMware, Inc.