Use Dell OpenManage IT Assistant to monitor the performance and power consumption of a single system or a group of systems on your network.
Performance Monitoring
Performance Monitoring helps you monitor the performance of a group of devices with supported Microsoft® Windows® or Linux operating systems over a specified period of time. Performance is monitored with the help of a set of performance counters available for each component. You can select and monitor these performance counters. You can configure thresholds for each performance counter and also configure alerts to be sent when the thresholds are crossed.
Using the Performance Monitoring feature, you can view the performance of individual devices rolled up on the Device tree. The overall performance of a device is calculated as the worst case status of the individual performance counter attributes monitored for the device. For example, if the status for the CPU Utilization counter is critical and the status of the memory paging counter is warning, the overall performance status of the devices is displayed as critical. From the Device tree, you can drill down to the performance counters and take appropriate actions.
To view details of how each performance counter is performing on a Dell system, do the following:
From the Device tree, expand the Server category
Select the system for which you want information.
On the right hand side pane, select the Performance and Power tab.
This tab displays the performance and power counter information for the selected system.
From this view, you can create multiple tasks to monitor multiple devices and manage these tasks, view results, and logs of these tasks.
NOTE: Performance monitoring enables you to monitor the usage of your systems
as against monitoring the health of the systems, which is provided by alerts and
notifications.
Power Monitoring
The power monitoring feature helps you to collect, store, and display the instantaneous values of power (watts) consumed, amperes drawn by each power supply, and the total energy consumed by a device.
NOTE: The power monitoring feature requires that the Dell OpenManage Server
Administrator version 5.3 be installed on the managed systems.
You can choose the appropriate power monitoring counters from the performance and power manager task wizard and select the frequency in which the data is collected. The collected data is available on the Performance and Power tab associated with each device, or the Execution Results tab associated with each Performance and Power Monitoring task.
The maximum value observed for a given system (Watts/Amps) is also collected during each polling cycle. This value is compared with the existing peak value in the IT Assistant database, and if the values are different, the database value is replaced with the values in the current polling cycle. The new value is displayed on the Group Summary and Maximum Values tab in the Performance and Power Monitoring screen.
NOTE: IT Assistant polls the managed systems at a frequency that you determine. If
you select a polling frequency that is too low, it is possible that the variations in the
power consumption are not captured adequately, and this may result in inaccurate
instantaneous power consumption graphs.
Performance and Power Monitoring in Tom's Enterprise-Size Business
Tom wants to use this feature to monitor how the Dell systems, specifically the PowerEdge x9xx systems, on his network are being used.
His main considerations for using this feature are:
Are the systems on my network under- or over-utilized?
Do I need to move my hardware (for example, CPU) or applications to another system?
How are my systems performing during peak and non-peak hours?
What is the energy consumption and the peak power values in my systems?
Would I need to balance the load among my systems?
To be able to answer these questions, Tom would need to perform the following:
Create a performance and power monitoring task
Monitor the systems over a period of time
View the data on the IT Assistant console
Export the data into comma-separated values and save it for later use
Creating a Performance and Power Monitoring Task
To create a performance and power monitoring task, Tom performs the following steps:
Tom selects Manage→ Performance and Power Monitoring from the
menu bar.
Tom right-clicks Performance and Power Monitoring Task and selects
New Task....
The New Task Wizard appears.
Tom enters a descriptive name for the task. For example, All x9xx systems.
Tom clicks Next.
Under Select Schedule, Tom selects a start date and an optional end date
to measure the performance attribute. He selects a 24-hour schedule to
monitor his systems during peak and non-peak hours.
Tom also adjusts the sampling interval based on how often he wants to sample his systems.
NOTE: Tom should take the network capacity into consideration. A bigger
sampling interval would not give an accurate picture and a smaller interval
may load the network and the monitored systems.
NOTE: The minimum frequency that Tom can set is two minutes, which means
that the task will be triggered every two minutes.
Under Select Attributes, Tom selects the performance, as well as the power
monitoring counters: CPU and Memory attributes (for performance), Power
Consumption, Energy Consumption, and Peak Amperage (for power
management). He sets their warning and critical threshold values and
specifies the number of samples for which the threshold values should be
crossed. For example, he sets the warning threshold for:
%Kernel Utilization Time as > 70% for 10 samples
%Processor Utilization Time as > 70% for 10 samples
%Power Consumption as > 1000 W
%Amperage per Power Supply as >7000 milliAmps for 10 samples
NOTE: The Power Monitoring attributes are supported only on limited Dell
systems.
NOTE: Tom cannot set threshold values for energy and peak measurement
(Peak power and Peak amperage) counters.
And the critical threshold for:
%Kernel Utilization Time as > 90% for 15 samples
%Processor Utilization Time as > 90% for 15 samples
%Power Consumption as > 1200 W
%Amperage per Power Supply as > 10000 milliAmps for 10 samples
Tom can select Send Warning Alert or Send Critical Alert to receive warning or critical alerts in the Alert logs.
NOTE: If Tom sets a smaller sampling interval but selects a large number of
counters, and devices, he may see a warning message indicating excess
resource utilization. Tom should set a higher sampling interval or decrease the
number of counters and devices to avoid this situation.
Under Device Selection, Tom can select the groups having Dell x9xx systems
from the tree or provide a query.
Under Enter Credentials, Tom enters the operating system User ID and
Password, which is valid for all selected devices.
Tom reviews his selection in the Summary screen and clicks Finish.
The All x9xx systems task appears on the tree under the Performance and Power Monitoring Tasks parent node.
NOTE: Performance monitoring tasks are not supported on virtualization hosts and
virtual machines. However, power monitoring tasks are supported on virtualization
hosts.
Monitoring the Usage of the Systems on the Network
To monitor the usage of all PowerEdge x9xx systems on the network, Tom performs the following steps:
Tom clicks the All x9xx systems task under the Performance and Power
Monitoring Tasks parent node.
The summary of the task is displayed under the Summary tab on the right
hand side of the screen.
To view the monitoring in greater detail, Tom selects the Execution
Results tab.
The counters keep a count of how a system is utilized.
Tom can sort on the counters to view how a particular component, for example, the %Kernel Utilization Time for each system is being utilized.
If the counters have been appropriately set, the counter colors would fairly indicate how well that systems are being utilized. Hover the mouse over the counter to get an indication of how the system component is performing.
For example,
If the counter is green for most of the time, it could indicate that the counter is well within the specified limits and could indicate that the system component can take more load, depending on the levels that Tom has set.
If the counter is yellow or red for a small amount of time, it could indicate that the system component is still partially-utilized.
If the counter is red for most samples, it could indicate that the system component is over-utilized.
See Table 7-1 for a sample of how systems on Tom's network may be utilized.
Table 7-1. Sample of Tom's network utilization
CPU Utilization
Memory Utilization
Network Usage
System 1
High
Low
Medium
System 2
Low
High
Medium
If %CPU Utilization Time is red for most of the samples collected (highly used), it could mean that some application is over-utilizing the CPU. Tom could consider moving this application to a system for which the %CPU Utilization Time is green for most samples. In this case, from System 1 to System 2. Tom could also move a memory module from System 1 to System 2 to balance the load, or he could consider upgrading the hardware or purchasing new memory modules.
If Tom monitors his systems over the weekend, and finds out the network and CPU utilization counters are green (within the specified range) for 70% of the samples, yellow (non-critical) for 20% of the samples, and red (critical) for 10% of the samples collected, it could mean that the network and CPU utilization counters could be red for most samples during the weekdays. The systems will be overloaded, and Tom could decide to add more systems to his network or decide on some other form of load-balancing.
Figure 7-1. Sample Performance and Power Monitoring Screen
In the Execution Results tab, Tom can right-click a counter and do one of
the following:
Select Export. This will export the details into a comma-separated values (CSV) file. Tom can then use other tools like Microsoft® Excel® to generate better reports.
Click View Chart to view the graphical representation of the performance, aggregate power consumed, and the aggregate energy consumed information of the device. Tom can give a time range and view the system usage graphically.
NOTE: Tom can also view the charts and export them from the Summary
tab, in the lower pane.
Click Delete Execution Results.
Right-click a column header and select
Customize View. This view customizes the view for the devices.
In the Group Summary and Maximum Values tab, Tom can view the
maximum watts/amps and the aggregate power or energy used by each
device for which this task is running.
NOTE: The Group Summary and Maximum Values tab is available only if Tom
selects the aggregate (Aggregate power and Aggregate energy) or peak
counters (Peak power and Peak amperage) in the Power Management
attribute.
In the Execution Log tab, Tom can view the execution summary
information for each run of the task. He can also use the time selection
fields to select the From time he wants to view the logs.
NOTE: Execution log entries older than 14 days will be purged.
In the Performance and Power tab on the Device tree, Tom can view the
performance and power counters information for the selected device.
Suggested Threshold Configuration for Performance and Power
Monitoring
Table 7-2 shows the sample threshold settings for each performance and power counter.
Table 7-2. Sample Threshold Settings for Performance and Power Counters
Resource
Performance Counter Attribute
Suggested Threshold
Comments
CPU
%Processor Utilization Time
Less than 85%
Total processor usage should remain under 85%, infrequent spikes exceeding 85% for brief periods is acceptable.
System
Context Switch/second
Depends on the system activity
Continued spikes for a prolonged time may indicate an increase in system load.
System
Processor Queue Length
2
Depends on the number of processors in the system. This is an instantaneous number. Needs observation over several cycles.
Memory
Available Memory
Less than 10 -20% of installed RAM Less than 4MB for systems with large memory
If available memory is under 10% 20% of the installed RAM for an extended period, it may indicate need for more memory.
Memory
Pages/Second
Less than 20
Should remain under 20 with the exception of brief spikes.
Memory
%Page File Usage
95%
Review this value in conjunction with Available Memory and Pages/Second.
Sharp deviation from average values for an extended period of time. Depends on the type of network
A sharp increase or decrease above normal levels is a strong indicator of network issues.
Physical Disk
Physical Disk I/O per Second
Depends on manufacturer's specifications
Check the specified transfer rate for your disks to verify that this rate does not exceed the specifications. In general, Ultra Wide SCSI disks can handle 50 to 70 I/O operations per second.
Logical Disk
Free Space
Less than 15%
Threshold value is relative to the total amount of disk space and the average I/O activity on the system.
Resource Usage by SQL Server and IT Assistant
Table 7-3 shows the recommended hardware configuration required for performance and power monitoring.
Table 7-3. Recommended Hardware Configuration for IT Assistant for Performance and Power Monitoring
Minimum Number of CPUs
Minimum Memory Required
Database
Maximum Number of user sessions per user
Maximum Number of Performance Counters
Minimum Supported Sampling Frequency
Maximum Number of Devices
Single CPU 2.0 GHz
512 MB
MSDE/SQL Express 2005
1
10
2 minutes
15
Single CPU 2.0 GHz
512 MB
MSDE/SQL Express 2005
1
18
2 minutes
8
Single CPU 2.0 GHz
1 GB
SQL 2000/ SQL 2005 Server
2
10
2 minutes
30
Single CPU 2.0 GHz
1 GB
SQL 2000/ SQL 2005 Server
2
18
2 minutes
20
Dual CPU 2.0 GHz
1 GB
SQL 2000/ SQL 2005 Server
2
10
3 minutes
100
Dual CPU 2.0 GHz
1 GB
SQL 2000/ SQL 2005 Server Enterprise Edition
5
10
5 minutes
200
NOTE: The hardware configuration listed in this table refer to the minimum
supported configuration. For the most recent update on these requirements, see the
IT Assistant readme on the Dell Support website at support.dell.com.