Product Documentation

Performance overview

Jun 26, 2017

The Citrix SCOM Management Pack for NetScaler consists of two parts:

  • A collection of SCOM management packs that are imported into SCOM (the server-side part)
  • The Citrix SCOM Management Pack for NetScaler Agent

The server-side part

Management packs included in this part are collections of discoveries, monitors, rules, and tasks for the Citrix NetScaler. This part does not add to the basic resource requirements(CPU and memory) of the SCOM management server where they are imported.

Citrix SCOM Management Pack for NetScaler Agent

Both, the performance and resource consumption of the Citrix SCOM Management Pack for NetScaler Agent depend mainly on the size of your NetScaler environment-specifically on the number of discovered and monitored NetScaler objects.

Configuration specifications

All the performance numbers in this document are valid for environments that:

  • are monitored with the specified product version of Citrix SCOM Management Pack for NetScaler
  • match the documented configuration specifications for NetScaler and SCOM
  • use the default configuration of management packs in terms of which rules and monitors are enabled (this applies to management packs included in NetScaler Management Pack and management packs bundled with SCOM)
  • use the default configuration of SCOM management servers and SCOM agents, without fine-tuning or making any special adjustments

Note: Factors such as different hardware specifications and conditions of your environment might cause discrepancies between the values you observe and ones documented here.

Validated Citrix SCOM Management Pack for NetScaler version 

The product version of Citrix SCOM Management Pack for NetScaler listed in the following table was validated:

Product Version
Citrix SCOM Management Pack for NetScaler
1.17.93

Citrix NetScaler configuration specification

Specification item Value

Software version

Citrix NetScaler 11.0 and 10.5

Virtualization platform

XenServer 6.5

Instantiated packet engines (NetScaler 11.0)

1

Instantiated packet engines (NetScaler 10.5)

1

Appliance type

NetScaler VPX

NetScaler devices in the monitored environment

2

Microsoft System Center Operations Manager configuration specification

With this configuration, the SCOM database and data warehouse servers are deployed outside the SCOM management server on a different machine.

Resource pool configuration
Specification item Value
SCOM resource pools 2
SCOM management servers in the resource pool 2

 

Computer: SCOM management server
Specification item Value
Compute Eight virtual CPUs; CPU clock speed of 2.67 GHz
Memory 16 GB of RAM
Software version Microsoft System Center Operations Manager 2012 R2

 

Computer: SCOM database and operation management server
Specification item Value
Compute Eight virtual CPUs; CPU clock speed of 2.67 GHz
Memory 24 GB of RAM
Software version Microsoft SQL Server 2014

 

Computer: SCOM database and data warehouse server
Specification item Value
Compute Eight virtual CPUs; CPU clock speed of 2.67 GHz
Memory 24 GB of RAM
Software version Microsoft SQL Server 2014

Monitoring ability

The following table lists the lab setup in which Citrix SCOM Management Pack for NetScaler was successfully validated with the specified NetScaler and SCOM configurations. NITRO API was used for monitoring. Data was gathered at the following intervals: 15 minutes for rules, 5 minutes for monitors, and 4 hours for object discovery.

Maximum number of monitored objects (for the specified configuration)

Item

Data Set 1

Data Set 2

Data Set 3

NetScaler objects discovered and monitored by Citrix SCOM Management Pack for NetScaler1

app. 6,000

app. 10,000

app. 14,000

1 Refers to the total number of objects of any type in any validated environment (with two appliances). For examples of object type distribution, see the following table.

Object type distribution examples (with load balancing)

Object type2

Data Set 1 

Data Set 2

Data Set 3 

Appliances

2

2

2

LoadBalancing Service

2,001

2,001

4,905

LoadBalancing ServiceGroups

1,751

4,001

4,951

LoadBalancing VirtualServer

2,202

4,202

5,053

System MemoryFeature

65

65

35

Other object types3

60

60

30

2 This table lists examples of object types that are monitored by Citrix SCOM Management Pack for NetScaler. Object type distribution examples Data Set 1, Data Set2, and Data Set3 explain the calculation of the maximum number of monitored objects.

3 Examples of other object types include Features, Global Settings, HTTP Settings, Interface, IPv4, Licenses, Memory Pool, Modes, Other Settings, SSL Certificate, TCP Settings, Timeout Settings, Virtual LAN, Other Settings, and so on.

Resource consumption

The resource consumption of the product was measured on the validation sets. Windows Performance Monitor was used as the measuring tool. 

Average computer and memory consumption of the agents 

The average computer and memory consumption of the agents was measured over a 24-hour period. Based on the measurement results, the average percentage of processor time and average memory usage of the MPNSMonitorSvc, HealthService, and MonitoringHost services were determined. 

Average consumption on a SCOM resource pool member

Citrix SCOM MP for NetScaler

Data Set 1 

Data Set 2 

Data Set 3

SCOM Server 1

SCOM Server 2

SCOM Server 1

SCOM Server 2

SCOM Server 1

SCOM Server 2

MPNSMonitorSvc percentage of processor time (per CPU)4

0.08% 

0.12%

0.12%

0.15%

0.17%

0.20%

HealthService percentage of processor time (per CPU)4

0.09% 

0.10%

0.20%

0.22%

0.25%

0.90%

MonitoringHost percentage of processor time (per CPU)4

3.31%

3.20%

9.27%

8.82%

16.62%

15.56%

MPNSMonitorSvc memory usage

366 MB

753 MB

475 MB

791 MB

826 MB

1,192 MB

HealthService memory usage

392 MB

403 MB

687 MB

671 MB

734 MB

1,124 MB

MonitoringHost memory usage

993 MB

1,025 MB

1,386 MB

1,358 MB

1,740 MB

1,780 MB

4CPU usage is calculated as the quotient of the % Processor Time counter and the number of processor cores or virtual CPUs.

The following figures show changes in CPU usage of HealthServiceMonitoringHost, and MPNSMonitorSvc after these services were started for the first time. Values were taken from the first part of the entire measurement time period (first four hours) when the initial high data inflow loaded the computer resource substantially.

Figure Percentage of processor time used by HealthService, MonitoringHost, and MPNSMonitorSvc over time, measured for the validation set with approximately 6,000 objects on SCOM Server 1. The readings were similar on SCOM Server 2.

localized image

Figure Percentage of processor time used by HealthService, MonitoringHost, and MPNSMonitorSvc over time, measured for the validation set with approximately 10,000 objects on SCOM Server 1. The readings were similar on SCOM Server 2.

localized image

Figure Percentage of processor time used by HealthService, MonitoringHost, and MPNSMonitorSvc over time, measured for the validation set with approximately 14,000 objects on SCOM Server 1. The readings were similar on SCOM Server 2.

localized image

The CPU usage percentage for the MonitoringHost process grows linearly with an increase in the total number of monitored objects.

The following figures show typical patterns of changes in CPU usage of HealthServiceMonitoringHost, and MPNSMonitorSvc during a major part of the measurement time period after the initial high load had subsided.

Figure Percentage of processor time used by HealthService, MonitoringHost, and MPNSMonitorSvc over time, measured for the validation set with approximately 6,000 objects on SCOM Server 1. The readings were similar on SCOM Server 2.

localized image

Figure Percentage of processor time used by HealthService, MonitoringHost, and MPNSMonitorSvc over time, measured for the validation set with approximately 10,000 objects on SCOM Server 1. The readings were similar on SCOM Server 2.

localized image

Figure Percentage of processor time used by HealthService, MonitoringHost, and MPNSMonitorSvc over time, measured for the validation set that has approximately 14,000 objects on SCOM Server 1. The readings were similar on SCOM Server 2.

localized image

After completion of the initial discovery cycle, the CPU usage is steady and the peaks are in sync with the SCOM data collection interval.

The following figures show memory usage of HealthServiceMonitoringHost, and MPNSMonitorSvc over a 24-hour time period, measured with the three different validation sets on two SCOM Servers. The figures show that for a data set that has approximately 6,000 objects, the SCOM management servers must have at least 3.6 GB of physical memory available for the memory requirements of HealthServiceMonitoringHost, and MPNSMonitorSvc.

Figure SCOM Server 1 memory usage with approximately 6,000 objects 

localized image

Figure  SCOM Server 2 memory usage with approximately 6,000 objects

localized image

The following figures show that with a data set of approximately 10,000 objects, the SCOM management servers must contain approximately 3.8 GB of physical memory for the memory requirements of HealthServiceMonitoringHost, and MPNSMonitorSvc

Figure SCOM Server 1 memory usage for approximately 10,000 objects

localized image

Figure SCOM Server 2 memory usage for approximately 10,000 objects

localized image

The following figures show that with a data set having approximately 14,000 objects, the SCOM management servers must contain at least 5.7 GB of physical memory available for the memory requirements of HealthServiceMonitoringHost, and MPNSMonitorSvc.

Figure SCOM Server 1 memory usage for approximately 14,000 objects

localized image

Figure SCOM Server 2 memory usage for approximately 14,000 objects

localized image

Storage consumption of the SCOM database server
The storage consumption of the SCOM database server was measured in an environment with two NetScaler VPX appliances. The measurement process spanned a period of 30 days, during which the NetScaler objects were added to the monitored environment at different intervals of one or three days. Based on the results, the maximum consumption of the SCOM database (OperationsManager) and the consumption growth of the SCOM data warehouse (OperationsManagerDW) were determined.

Maximum consumption of the SCOM database

SQL Server database

Database file name

Monitored objects

Maximum storage consumption

OperationsManager

MOM_DATA

2,600

0.25 GB (237 MB)

6,000

0.42 GB (433 MB)

9,800

0.97 GB (994 MB)

14,000

1.45 GB (1,483 MB)

Consumption growth of the SCOM data warehouse

SQL Server database

Database file name

Monitored objects

Consumption growth

Per day
(in MB)

Per week4
(in MB)

Per month4
(in GB)

OperationsManagerDW

MOM_DATA

2,600

29

203

0.85

6,000

49

344

1.44

9,800

86

601

2.51

14,000

197

1,377

5.76

4  Projection.

Figure Storage requirements of SCOM database (OperationsManager) over  time, measured with six different validation sets

localized image

In the figure above, you can see a strong correlation between the number of monitored NetScaler objects and the storage consumption of the SCOM database. Under stable conditions, the SCOM database storage consumption (mean value) increases linearly with the addition of new objects. The consumption stabilizes when you stop adding objects.

Figure Storage requirements of the SCOM data warehouse (OperationsManagerDW) over time, measured with six different validation sets

localized image

In the figure above, you can see a linear growth in the storage consumption of the SCOM data warehouse over the validation period. The storage consumption increases independent of the addition of new objects and persists after the object count remains constant. The increase though depends on the number of added objects.