Veryx™ and PktBlaster™ are trademarks of Veryx Technologies Pvt. Ltd. All other
trademarks of respective owners are acknowledged.
This document may not be copied either in whole or in part by any means or media
without the permission of Veryx Technologies Pvt. Ltd.
THE PROGRAMS FURNISHED ARE PROVIDED ON AN "AS IS" BASIS, WITHOUT ANY
WARRANTIES OR REPRESENTATIONS EXPRESS, IMPLIED, OR STATUTORY; INCLUDING,
WITHOUT LIMITATION, WARRANTIES OF QUALITY, PERFORMANCE,
NONINFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
NOR ARE THERE ANY WARRANTIES CREATED BY A COURSE OF DEALING, COURSE OF
PERFORMANCE, OR TRADE USAGE. LICENSOR DOES NOT WARRANT THAT THE
PRODUCTS WILL MEET LICENSEE'S NEEDS OR BE FREE FROM ERRORS, OR THAT THE
OPERATION OF THE PRODUCTS WILL BE UNINTERRUPTED.
Page 4
Table of Contents
1 About Pkt Blaster SDN Controller Test ..................................................... 1
Key Features .......................................................................................................................... 2
Working with PktBlaster SDN Controller Test ........................................................................ 2
User Interface ................................................................................................................. 2
PktBlaster SDN Controller Test is the industry’s first integrated test solution for
benchmarking the performance of an SDN controller as well as emulating SDN
networks in a single box. It supports a number of real-world application traffic for
performance testing and emulation. It runs either on a Veryx hardware platform or
on suitable commodity servers.
PktBlaster SDN Controller Test solution consists of two product components that
are licensed separately:
Network Emulator and
Performance Benchmarking
PktBlaster SDN Controller Test-Performance Benchmarking simulates up-to one
thousand switches and millions of nodes and thereby significantly reduces
dependency on large-scale physical network infrastructure for testing and
benchmarking SDN controllers.
Some of the performance benchmarking capabilities include:
Throughput and Robustness measurement – per second
Latency measurement – in milliseconds
Detailed Tx/Rx statistics of switch group network per iteration
Comprehensive reporting and analysis
PktBlaster SDN Controller Test – Network Emulator is available separately and it
supports real-time emulation of large-scale SDN networks. Some of the emulation
capabilities supported include:
Emulation of upto 100 SDN devices and upto 100 endpoints with millions of
flows
Generation of multiple traffic flows from a single simulated port.
Supports network flow path tracing and network failure simulation.
Detailed traffic and flow statistics.
Page| 11
Page 9
2
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
For more details, please refer to the PktBlaster SDN Controller Test Network
Emulator User Guide.
Key Features
The following are some of the key features of PktBlaster SDN Controller Test for
performance benchmarking:
Measurement of Performance Throughput, Latency and Robustness:
For OpenFlow versions 1.0, 1.3 or both
Uses TCP connection mode
Programmatic Support (REST API)
Comprehensive Reporting and Analysis
Working with PktBlaster SDN Controller Test
PktBlaster SDN Controller Test runs either on Veryx Hardware or on standard
commodity servers. The platform allows easy configuration of controllers and
supports easy web-based access.
User Interface
PktBlaster SDN Controller Test supports user friendly GUI for configuration and
testing. Also, PktBlaster SDN Controller GUI supports creation of various real-world
application data for testing the performance and functionality.
Troubleshooting
PktBlaster SDN Controller Test provides extensive debugging options at multiple
levels. It provides detailed statistics of the requests and responses for a test run for
quick identification of problem root cause.
PktBlaster SDN Controller Test report provides comprehensive insight on controller
metrics such as flow set-up rate and flow set-up time in the form of graphs.
The individual test report provides OF connection time and detailed statistics of
switch group network.
Figure 1: PktBlaster SDN Controller Test-Performance Benchmarking Deployment
Page 11
4
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
View the status of controller tests
that were performed in the
previous week.
Controllers
Add new controller information.
Modify and delete existing
controller information.
Traffic Profile
Add a new traffic profile.
Modify and delete an existing
traffic profile.
Emulation
Execute Emulation tests.
Performance
Configure and execute
Performance tests.
In a new installation, no data is displayed in the Dashboard. After the tool has been
used to test controllers, a dashboard showing the results of testing is displayed (as
shown in the View the Dashboard section).
About the Home Screen
The tabs in the PktBlaster SDN Controller Test allow testers to configure and
perform tasks related to setting up and executing tests. The following table shows
the tabs that you will see in the Home screen (Figure 3). To learn more about a
module, click on the corresponding link.
This chapter describes how you can configure a SDN controller in PktBlaster SDN
Controller Test.
An SDN controller can be tested for performance throughput, latency and
robustness. To perform these tests, you need to add the controller to the tool and
the controller parameters need to be configured.
To test the performance of a controller,
The controller needs to be connected to the network
The controller should be reachable through the port of the tool
To configure the parameters of a controller, click the Controllers tab. The
Controllers screen is now displayed.
In the Controllers screen, you can:
View the details of the controllers that have been added
Add the profile of a new controller
Page| 1111
Figure 6: Controller Screen
Page 19
12
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
The version number that relates to the date and time
of the build (in this case the build version of the
controller that is tested).
IP Address
The IP Address of the server where the controller is
installed (the controller runs its applications on this
server).
Controller Port
The port number at which the controller listens. The
default port number is 6633. This can be set to a
maximum value of 65535.
Mode of Connection
The mode of connectivity between the switches and
the controller. The TCP connection mode is
supported.
Open Flow Version
The version number of the OpenFlow protocol
(protocol used by the SDN controller to communicate
with the switches) that is supported by the controller.
The protocol versions supported are 1.0, 1.3 and both
versions (1.0, 1.3).
Table 2: Add a New Controller Fields
All fields that have an beside them are mandatory.
2. After you have entered the relevant information, click to save the
controller details
The new controller is now added to the tool as shown in Figure 8.
Page 21
14
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
To view the details of a controller, point to the controller with your cursor. The
details of the controller are now displayed in a pop-up as shown in the Available Controllers screen (Figure 9).
Key features to Customize Controller View
The tabular information relating to controller configuration is displayed as shown in
the Controller Configuration Screen.
Page 23
16
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
In the Controller Configuration screen, you can customize the information that is
displayed, using the features shown in the Controller Configuration-CustomizeDisplay screen.
The information can be customized based on the following:
Grouping: Enables you to group the information based on a specific field.
Customizable Columns: Enables you to select the fields that you want, to be
displayed.
Edit SDN Controller Details
To modify the parameters of a controller that has been added,
1. Click the Controllers tab to display the Available Controllers screen (Figure
9).
2. From the list of controllers that is displayed, select the controller that you
want to modify. The details of the selected controller are displayed in the
relevant fields as shown in the Controller Details screen (Figure 12).
Page 25
18
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
This chapter describes how you can configure the type of traffic that needs to be
sent to the controller. Predefined traffic types such as ARP_Reply, ARP_Request,
TCP, and UDP are available in Traffic profile module. It is possible to select or modify
the existing traffic types and create new traffic types.
These traffic types are used for the packet-in messages that are sent to the
controller. After a traffic profile is created, it can be used to test the performance of
a controller.
To configure the traffic that needs to be generated, click the Traffic Profile tab. The
Traffic Profile Screen displays the predefined traffic types ARP_Reply, ARP_Request,
TCP and UDP.
Figure 15: Traffic Profile Screen
Page| 2121
Page 29
22
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
In the Traffic Profile screen, you can view the existing profiles that have been
defined. When you click on a traffic profile, you can view the field details and the
modifier details for that profile.
In the Traffic Profile screen, you can:
Add a new traffic profile
View traffic profiles
Modify an existing traffic profile
Delete a traffic profile
Add a New Traffic Profile
To define a new traffic profile:
1. In the Traffic Profile screen, click . The Add a New Traffic Profile screen
The starting position in the field from which the mask is applied. This
indicates the position (starting from 1) from which the modifier is
applied.
Mask
The value corresponding to the number of byte(s) that are modified and
the offset at which the modifier is applied. For example, when 1 byte of
a field (of length 4 bytes) is modified, the mask value that is applied
from an offset position 3 is 00 00 ff 00.
Count
The number of combinations that are generated with a specified mask
value. For example, when 1 byte of a field is modified, the maximum
number of unique values that will be generated is 255.
In the Add a New Traffic Profile screen, you can view the following:
The field details for the packets.
The modifier information for the relevant fields.
In the Add a New Traffic Profile, screen (Figure 16):
1. In the Profile Name text box, enter a suitable name that uniquely identifies
the traffic type associated with the profile. For example, to define a Layer 3
custom traffic, you could use a profile name such as L3_CUSTOM_1.
2. In the Packet Length box, type or select the length of each packet in the
traffic profile. The length of a packet (includes the header and data parts) can
vary from a minimum of 64 bytes to a maximum of 512 bytes.
In the Field Details section:
The values in the fields can be modified by means of the parameters shown in the
following table.
Table 3: Fields and Associated Modifiers
Page 31
24
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
3. In the Destination text box, the default physical address of the packet
destination is displayed. This is a 6-byte address where you can modify the
last 4 bytes.
4. In the Source text box, the default physical address of the packet’s source is
displayed. This is a 6-byte address where you can modify the last 3 bytes.
Although you are allowed to modify only the last 3 bytes in the Source field,
the module allows you to apply a modifier from the third offset. However,
when the traffic profile with such a modifier is saved, a warning is displayed.
From the Type drop-down list, select the protocol that is used to transmit the
packets in this traffic profile. Select the suitable Type for traffic that needs to
be generated:
The following sections explain the fields that are required to be set when the
Type is IPV4 or L2 custom traffic.
Page 35
28
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
When the next protocol of IPV4 is set to TCP or UDP, the Traffic Profile-TCP/UDP
screen is displayed.
Figure 18: Traffic Profile-TCP/UDP
To specify the ports for a TCP or UDP protocol:
1. In the Destination Port box, type or select the port number that identifies
the type of the packet. For example, a packet with port 80 is directed to be
processed by a Web server. This is 2 bytes long and you can modify all the
bytes of the destination port. The destination port number can be assigned a
value between 1 and 65535.
2. In the Source Port box, type or select the port number that is used in
responses after the packet has reached the destination port. This is 2 bytes
long and you can modify all the bytes of the source port number. The source
port number can be assigned a value between 1 and 65535.
3. Click to save the traffic profile.
Page 37
30
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
The right pane displays all the traffic profiles that have been created along with the
predefined traffic types. When you click on a specific traffic profile, the details of the
traffic profile are displayed in the left pane.
Modify a Traffic Profile
To modify the information pertaining to a traffic profile:
Click the Traffic Profile tab. The Traffic Profile screen (Figure 15) is displayed.
2. In the right pane, click the traffic profile that you need to modify to display
the details in the left pane.
3. Click to display the details of the selected traffic profile in editable
mode as shown in the Edit a Traffic Profile screen.
This chapter describes the performance tests that can be executed on a controller. It
also describes how you can configure the test parameters to execute a test.
Types of Tests
When a controller is added and connected to the network, it can be tested for the
following:
Throughput is the total number of flows generated per second by the controller or
the rate at which flows are generated by the controller.
To test the Performance Throughput of a controller, the switches of the participating
switch group network send out as many packet-in messages towards the controller
as possible, in a given duration.
The throughput is measured based on the Flow Measured by parameter that is
specified by the user. The user can choose to measure the Packet out messages or
Flow Mod messages that the controller sends back as a response to the switches.
The rate of flow insertion of this measurement is the throughput.
Performance Latency
This is the time taken by the controller to create a flow entry in the flow table. It is
also known as the Flow Insertion time. The Flow Insertion time can also be
described as the time taken to obtain a response from the controller.
Performance Robustness
To test performance robustness, valid packet-in messages are sent to the controller
along with malformed packet-in messages. The controller should handle and
Page| 3535
Page 43
36
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
respond to the valid packet-in messages and ignore the malformed packet-in
messages. Controller response is measured either in Packet out messages or in Flow
Mod messages.
The test is divided into 4 iterations:
For the first 25 % of the test duration, normal packet-in messages are sent as
test traffic to stabilize and baseline controller performance.
In the second 25% of the test duration, error packets are sent along with
normal packet-in messages and the percentage and the error value depend
on the test configuration.
In the third 25% of the test duration, the percentage and the error value of
the error packets is doubled in comparison with the second iteration.
In the fourth iteration, normal packet-in messages are sent to verify the
stability of the controller performance.
Testing Modes
The Throughput and Latency tests can be carried out in different modes as shown in
the following table. To know more about a testing mode, click on the corresponding
link in the table.
This test loads the controller to the maximum and measures the total number of
Packet out messages or flow mod messages that the controller is able to send back
to the switches.
The test will run for the configured switches of the participating switch group
network. OpenFlow channels are established for the configured number of switches.
The tool will generate packet-in messages for each switch, for the test duration that
is specified.
This process is repeated for the number of iterations that was configured. During the
test, packet-in messages are sent to the controller so that the TCP window is always
full.
Throughput Test in Incremental Switch Mode
This test measures how the controller handles load when it is generated in
increments. In this mode, only one switch group can participate.
The test is executed in four iterations. The first iteration is executed for 25% of the
configured number of switches. OpenFlow sessions are then established for the 25%
of the configured number of switches. For subsequent iterations, the number of
switches is incremented in steps of 25% and OpenFlow sessions are established for a
further 25% of the switches until the total number of configured switches has been
included in the testing.
Latency Test in Normal Mode
This mode measures the OpenFlow controller’s request processing time under lowload conditions.
To run the latency test in normal mode, at least one switch group network has to be
selected. OpenFlow sessions to the controller are established for the total number
of configured switches of the participating switch group network.
A packet-in message is sent from each simulated switch to the controller. The
switches wait for a response from the controller before sending the next packet.
Thus, each simulated switch ensures that only a single outstanding new packet-in
message is sent to the controller.
Page 45
38
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
The simulated switch waits for a response (a Packet out message or a flow mod
message) before soliciting the next request. The total difference in milliseconds
between the time a packet-in message is sent and the response (Packet out or Flow
Mod message) is received. The latency computation is time stamp based. This test is
repeated for the number of iterations that have been configured.
Latency Test in Incremental Switch Mode
This mode measures the request processing time of an OpenFlow controller under
incremental switch load conditions. You can select only one switch group network
for a test.
The test is executed for four iterations. Each iteration, 25% of the switches
belonging to the participating switch group are added.
The first iteration is executed for 25% of the configured number of switches.
OpenFlow sessions are established for only 25% of the configured number of
switches. For subsequent iterations, the number of switches is incremented in steps
of 25% until the total number of configured switches has been included in the
testing.
Each simulated switch maintains exactly one outstanding new packet-in message,
waiting for a response before soliciting the next request. This process is repeated as
many times as possible within the configured test duration. The response (Packet
out message or Flow Mod message) received at the end of the time period is used to
compute the total difference in milliseconds that it takes for the controller to
process each packet-in message.
Configure Test Parameters
To execute a performance test on a controller, you need to configure the
parameters required for testing.
To configure and execute a test, click the Performance tab to display the
Performance screen.
The IP Address drop-down list shows the controllers that have been added. You can
select the controller for which the test is to be performed from the list. To view the
port number at which the controller is listening and other details see the Select Controller pane (Figure 26).
Step 2: Network Configuration
The Network Configuration enables you to do the following:
Create between 1 to 4 switch network configurations.
Each switch network configuration can have different versions of the
OpenFlow switch specification.
The maximum number of switches in all the switch group network
1. In Switch Group Name, specify the name of the simulated SDN network and
Click …
2. In OpenFlow Version, specify the version of the Open flow channel between
switches and controller.
3. In # of Switches, enter the number of OpenFlow switches that need to be
simulated.
For Throughput Tests and Latency Tests the maximum # of switches is 1000. For
Robustness Tests the maximum # of switches is 64.
4. In Start MAC address, specify the address of the first switch for each switch
group. For example:
If the MAC address of the first switch is 00:00:00:00:00:01 and consists of
10 switches in the first switch group, the first switch number of the 2nd
switch group Start MAC address will be 00:00:00:00:00:11.
If you specify a value less than 11 in the starting MAC address for the 2
nd
switch group, it will result in an error message.
5. In Number of Ports, specify the number of ports for each switch of the
simulated network.
6. In the Traffic Profile field, click Select beside the text box that allows you
select a suitable traffic profile carried by packet-in messages.
For details on how to create a Traffic Profile, refer to the Configuring a Traffic Profile chapter.
7. # of flows displays the number of hosts simulated in each switch.
Show Advanced Configuration
The components displayed in Figure 28 enable detailed configuration for each switch
in a switch group network. To view the range, let your cursor hover over the specific
field.
Page 49
42
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
#of Meter Tables: Specify the number of meter tables.
Max entries for Flow Table: Specify the number of entries in the flow table.
Switch Buffer Size: Configure the buffer size. Zero indicates there is no buffer
available.
Step 3: Test Configuration
In the Test Configuration pane, you can specify the type of test to be performed, the
testing mode and other test parameters.
Figure 30: Test Configuration Pane
In the Test Configuration pane (Figure 30):
1. To specify the Test Type you want to perform, select Throughput, Latency or
Robustness as applicable.
2. From the Test Mode drop-down list, select the required mode of testing.
Refer to the Testing Modes section for more details on the different modes used in testing.
This is not applicable for Robustness Test.
Page 51
44
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
3. In Flow Measurement By, select the response used to compute the
Throughput, Latency or Robustness of the controller. This could be Packet
out message or Flow Mod message.
4. The Duration box, displays the duration of the performance test in seconds.
5. In the # of Iterations box, type or select the number of times the test has to
be repeated.
For example, if the duration of the test is set to 5 seconds and the number of
iterations is 5, then there would be 5 executions of the test, each for 5
seconds.
You can set the Duration of the test to a maximum of 60 seconds. The maximum number of
iterations for a test is 60. For Incremental Mode Test, the number of iterations is set to 4.
This step is not applicable for Robustness Test.
6. In the Iteration Delay (sec) parameter, specify the number of seconds the
test should pause before the next iteration starts.
7. In the Warmup Time (sec) parameter, specify the number of seconds the test
should wait before sending the test traffic to the controller.
8. In the Malform Field, you can select the packet-in messages fields that can
be malformed.
OF Version: OF version in the header.
OF Header Length: Total length of the packet-in message in the OF
header.
OF Message Type: Type of message in the OF header.
9. In Malform Pkts%, specify the percentage of malformed packet-in messages
that will be sent along with normal packet-in messages. The error value
depends on the Malform field.
Malform Field and Malform Pkts % parameters are applicable only for Robustness Test.
This chapter describes the results of a test execution and the procedure to reexecute a test. The details explained in this chapter apply to Throughput tests,
Latency tests and Robustness tests.
Refer to Chapter 7: Executing Throughput Tests, Chapter 8: Executing Latency Tests and
Chapter 9: Executing Robustness Tests for details specific to these test executions.
After you have configured all the parameters and started the execution of the
required test, you can view the progress of the test execution.
View a Test in Progress
The progress of Test Execution for a Throughput test in Normal mode are displayed
in the Test Execution in Progress(Upper and Lower) screens.
Figure 31: Test Execution in Progress (Upper)
Page 55
48
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
When a test execution completes successfully, the response of the controller is
captured and plotted in a graph and in a table. The test results display the following
information about the test that was performed:
The configuration details for the test that was executed
A graph that shows the output trend
A table that shows precise details
This information is represented in the figures that follow. (Figure 33 to Figure 35)
The test result screens (Figure 33 to Figure 35) shows the throughput of the
controller over the configured number of iterations and highlight the following
information about the test execution:
For each iteration, the flow insertion rate is generated for the number of
responses received per second from all switches belonging to the switch
group network(s) is plotted.
A comparative plotting of the number of packet-in messages and the number
of responses generated by the controller per second.
The maximum, minimum, and average number of flow responses generated
by the controller per second for all the configured switches.
The table shows a comparison of the number of packet in messages and the
number of responses generated per second for all the switches belonging to
switch group network(s) that has been simulated.
Re-execute a Test
A test can be re-executed with any of the following:
Existing configuration
Modified test parameters
In a throughput test, the flow insertion rate of the Packet out messages or Flow Mod
messages are sent by the controller over a specified duration is measured. The
Packet out messages or Flow Mod messages (depending on the mode of
measurement specified by the user) are the response of the controller to the packetin messages sent from the simulated switches.
The values shown in the test executions are hypothetical and are used for the purposes of
explanation.
Throughput test can be executed in two modes:
Normal
Incremental Switch
Throughput Test in Normal Mode
This mode of testing loads the controller with the maximum number of packet-in
messages for a given time. The total number of Packet out messages or Flow Mod
messages that the controller is able to send back to the switches is then measured.
This process is repeated for the number of iterations that was configured.
To run a throughput test in the Normal mode:
1. Click the Performance tab to display the Performance screen (Figure 25).
2. Configure the Select Controller pane and Network Configuration pane
(Figure 27).
For details of how to configure the panes, refer to Configure Test Parameters section in the
Performance Tests chapter.
Page| 5353
Page 61
54
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
To continue configuring the test parameters, go to the Test Configuration pane.
Figure 37: Test Configuration-Throughput Test Mode
4. To specify the Test Type, select Throughput.
5. From the Test Mode drop-down list, select Normal.
6. In the Flow measurement by, the response sent by the controller can be
Packet out or Flow Mod message.
7. In Duration, specify the duration of the test measured in seconds.
8. In the # of Iterations box, type or enter the number of times the test is to be
repeated. Here, the number of iterations is set to 5.
For example, if the duration of the test is set to 5 seconds and the number of
iterations is 5, then there would be 5 executions of the test, each for 5
seconds.
You can set the duration of the test to a maximum of 60 seconds. The maximum number of
9. In Iteration Delay (sec), specify the time for the test to pause before starting
the next iteration.
10. In Warmup Time (sec), specify the number of seconds the test will wait after
the handshake with the controller is completed.
11. Click to run the test.
The Test Status for Performance Test-Throughput Normal Mode (Upper and Lower) screens (Figure 38 and Figure 39) shows test execution in Normal mode. The
results are measured based on Flow Mod messages.
Figure 38: Throughput Normal Mode (Upper)
Page 63
56
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
The result for this test execution is displayed as shown in the Throughput Normal
Result-Graph screen (Figure 40) and Throughput Normal Result-Table screen
Average OF Connection Time (ms): Average open flow connection time
taken for all the switches to establish a connection with the controller and
complete the handshake.
4. A comparative plot of the number of responses generated per second and
the number of packet-in messages. This highlights the performance trend of
the controller with respect to the number of response received.
5. The minimum, maximum, and average number of responses generated per
second is displayed.
Figure 41: Throughput Normal Result-Table
6. The table (Figure 41) shows a comparison of the number of packet-in
messages and the number of responses generated per second for all the
simulated switches of an individual switch group network for iteration 1. The
total OF connection time is also shown.
To download the test performance results displayed in the Switch Group
Statistics table, click CSV Group Statistics.
To download the details of test performance results for all switches in the
switch group network, click CSV Switch Statistics.
The test can be repeated with the same configuration or with a modified
configuration. For more information on executing the test again, refer to Re-
execute a test in the Test Execution Details chapter.
In this mode of throughput testing, OpenFlow packet-in messages are first sent to
the controller from 25% of the configured number of switches belonging to a single
switch group network. This test is repeated in increments of 25% until all the
configured switches have sent packet-in messages to the controller. This would
require 4 iterations. It is not possible to change the number of iterations in the test.
To run a throughput test in the Incremental Switch mode:
1. Click the Performance tab to display the Performance screen (Figure 25).
2. Configure the Select Controller pane and Network Configuration pane
(Figure 27).
For details of how to configure the panes, refer to Configure Test Parameters section in the
Performance Tests chapter.
To continue configuring the test parameters, go to the Test Configuration pane.
Figure 42: Test Configuration- Throughput Incremental Switch Mode
Page 67
60
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
4. From the Test Mode drop-down list, select Incremental Switch.
5. In Flow Measurement By, select the response used to compute the
Throughput, Latency or Robustness of the controller. This could be Packet out message or Flow Mod message.
6. In the Duration (sec) box, type or select the test duration. Here, the test
duration is set to 5 seconds.
The # of Iterations box is disabled in this mode as the test is set to repeat for 4 iterations.
7. In Iteration Delay (sec), specify the time for the test to pause before starting
the next iteration.
8. In Warmup Time (sec), specify the number of seconds the test will wait after
the handshake with the controller is completed.
9. Click to run the test.
The Test Status for Performance Test-Throughput IncrementalSwitch Mode (Upper and Lower) screens (Figure 43 and Figure 44) shows the test execution
progress for a throughput test in Incremental Switch mode. The performance is
measured based on Packet out messages.
The result for this test execution is displayed as shown in Throughput Incremental
Switch Result-Graph screen (Figure 45) and Throughput Incremental Switch ResultTable screen (Figure 46).
Page 69
62
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
The test result screens display the following information about the test that was
executed.
1. The X-axis in Figure 45 represents the number of switches from which,
packet-in messages are sent to the controller.
2. The Y-axis represents the number of Packet out messages generated by the
controller per second.
3. The flow insertion rate for the number of Packet out messages per second is
plotted for the number of switches on the X-axis. The rate of flow insertion
for the number of responses per second at a specific incremental stage is
displayed as a tooltip.
4. The minimum, maximum, and average number of responses generated per
6. Figure 46 displays the number of responses per second for each switch that
has been simulated along with the number of previous responses for the
participating switch group network.
To download the test performance results displayed in the Switch Group
Statistics table, click CSV Group Statistics.
To download the details of test performance results for all switches of
the switch group network click CSV Switch Statistics.
The test can be repeated with the same configuration or with a modified
configuration. For more information on executing the test again, refer to Re-execute
a test in the Test Execution Details chapter.
Page 71
64
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
Controller latency is the time taken to insert a flow into a switch. This is also known
as the Flow Insertion time and is measured in milliseconds.
It is possible to execute a Latency test in the following modes:
Normal
Incremental
Latency Test in Normal Mode
This mode of testing measures the time taken to receive a response from the
controller under low-load conditions.
During the test, the simulated switches of a participating switch group send a
packet-in message to the controller and wait for a response before sending the next
packet-in message. Latency is computed as the difference in time between the
sending of packet-in message and receiving of the response (packet-out or flowmod) calculated in milliseconds.
Latency measurements are time-stamp based and accurate. This process is repeated
for the configured number of iterations.
To run a latency test in the normal mode:
1. Click the Performance tab to display the Performance screen (Figure 25).
2. Configure the Select Controller pane and the Network Configuration pane
(Figure 27).
For details of how to configure the panes, refer to Configure Test Parameters section in the
Performance Tests chapter.
Page| 6565
Page 73
66
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
The Test Status for Performance Test- Latency Normal Mode (Upper and Lower)
screens (Figure 48 and Figure 49) shows the test execution progress for a latency
test in Normal mode. The performance is measured based on Packet out messages.
Figure 48: Latency Normal Mode (Upper)
Figure 49: Latency Normal Mode (Lower)
Page 75
68
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
The result for this test execution is displayed in the Latency Normal Result screen.
Figure 50: Latency Normal Result
The Latency Normal Result screen (Figure 50) and Latency Normal Table Report
(Figure 51) display the following information about the test that was executed.
1. The X-axis represents the number of iterations. (Here, number of iterations is
4).
2. The Y-axis represents the latency time in milliseconds.
3. The graph shows the average time taken to process a packet-in message in
each iteration. (Packet-in messages are sent from switches of configured
switch groups for 5 seconds). The average time taken to process a packet-in
message during a specific iteration is displayed as a tooltip.
OF Connection Time (ms): Time taken by all the switches to establish a
connection with the controller and to complete the handshake with the
controller.
Average OF Connection Time (ms): Average open flow connection time
taken for all the switches to establish a connection with the controller
and complete the handshake.
4. The minimum, maximum, and average time taken to process a packet-in
message is displayed.
Figure 51: Latency Normal Table Report
5. The Detailed Statistics table displays the number of iterations the test will
run. Each iteration shows the total time, measured in milliseconds, taken by
all the simulated switches to receive a response from the controller.
To download a copy of the Latency Normal Table Report click on the CSV
situated at the top right corner of the Detailed Statistics screen.
The latency test can be repeated with the same configuration or with a modified
configuration. For more information on repeating the test, refer to Re-execute a test
in the Test Execution Details chapter.
Latency Test in Incremental Switch Mode
This mode measures the time taken by the controller to process a packet-in message
under incremental switch load conditions. This is achieved by increasing the number
of switches part of a switch group that send packet-in messages to the controller in
steps of 25%.
Page 77
70
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
7. In the Duration (sec) box, type or select the test duration. Here, the test
duration is set to 5 seconds.
The # of Iterations box is disabled in this mode as the test is set to repeat for 4 iterations.
8. In the Iteration Delay (sec) box, specify the number of seconds the test
should pause before it starts the next iteration.
9. In Warmup Time (sec), specify the number of seconds the test should wait
after the handshake with the controller is completed.
10. Click to run the test.
The Test Status for Performance Test- Latency Incremental Switch Mode screen
(Figure 53 and Figure 54) shows the test execution progress for a latency test in
Incremental Switch mode of the participating switch group network. The responses
are measured based on Flow Mod messages.
The Latency Incremental Switch Result screen displays the following information
about the test that was executed:
1. The X-axis represents the number of switches. (Here, the participating switch
group network determines the number of switches).
2. The Y-axis represents the latency time in milliseconds.
3. The graph shows the average time taken to process a packet-in message
when the number of switches is incremented by 25%. Successive test
iterations will have the number of switches incremented in this manner. The
average time taken to process a packet-in message during a specific
incremental stage is displayed as a tooltip.
4. The minimum, maximum, and average time taken to process a packet-in
message is displayed.
Figure 56: Latency Incremental Switch Table Result
5. Figure 56 displays the number of switches for each iteration is computed in
multiples of 8 and the latency is measured in milliseconds.
To download a report of Incremental Switch Table Result, click CSV at the
top right of the detailed statistics screen.
The test can be repeated with the same configuration or with a modified
configuration. For more information on repeating the test, refer to Re-execute a test
in the Test Execution Details chapter.
Page 81
74
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
A Robustness test is done to monitor the behavior of the controller. Test
performance is measured depending on the response type. The response type and
number of malformed packet-in messages are configured using test parameters.
Valid packet-in messages along with malformed packet-in messages are sent to the
controller and the controller should handle and respond to valid packet-in messages.
There are no test modes in the Robustness Test.
To run a Robustness test:
1. Click the Performance tab to display the Performance screen (Figure 25)
2. Configure the Select Controller pane and the Network Configuration pane
(Figure 27).
For details of how to configure test parameters, refer to Configure Test Parameters section in
the Performance Tests chapter.
3. To configure the robustness test parameters, go to the Test Configuration
pane.
The Test Configuration procedures for Flow Mod messages and Packet out messages are
similar.
Page 83
76
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
10. In the Iteration Delay (sec) box, specify the number of seconds the test
should pause before it starts the next iteration to flush out the responses
received during the previous iteration.
11. In Warmup Time (sec), specify the number of seconds the test should wait
after the handshake with the controller is completed.
12. To run the test, select the switch group network in the network
configuration pane and click to run the test.
The Robustness test will run only for four iterations.
The Test Status for Performance Test- Robustness Flow Mod Result Report (Upper
and Lower) measures performance Robustness are displayed in Figure 58 and Figure
59.
Figure 58: Robustness Flow Mod Result Report (Upper)
Page 85
78
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
Figure 59: Robustness Flow Mod Result Report (Lower)
View Test Result
The test results for Packet-out messages are displayed in a manner similar to those
for Flow Mod messages. The result of this test execution is displayed in the
Robustness Flow Mod ResultReport (Lower) screen (Figure 59).
The Robustness Flow Mod Graph Result screen (Figure 60) and Robustness Flow Mod ResultTable Report (Figure 61) display the following information about the
test that was executed.
1. The X-axis represents the number of iterations. (Here, number of iterations is
set to 4).
2. The Y-axis represents the number of malformed packet-in messages along
with valid packet-in messages.
3. The graph displays the number of valid packet-in messages along with
malformed packet-in messages and Rx trend chart for the no of responses
received during each iteration.
4. The graph allows you to click on series that you do not want to display.
Figure 61: Result Table Report
5. The Result Table Report (Figure 61) displays the number of iterations the
test will run. The statistics provided for each iteration are: the percentage of
malformed packets, the number of malformed packets, and the number of
valid packets sent to the controller and the number of packet-out messages
received from the controller.
To download a copy of the Result Table Report click CSV situated at the topright corner of the Detailed Statistics screen.
The Robustness test can be repeated with the same configuration or with a modified
configuration. For more information on repeating the test, refer to Re-execute a test
in the Test Execution Details chapter.
Page 87
80
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
This chapter describes the reporting feature of PktBlaster SDN Controller Test. The
different reports that are generated provide comprehensive information about the
tests that were performed. The reports display results pertaining to the last one
week by default. However, you can change the period for which you want to view
results.
View the Dashboard
After PktBlaster SDN Controller Test has been used to test controllers, you will see
the dashboard when you log in to the tool.
The Dashboard (Figure 62) is displayed as a high-level view of test details in the
Home screen.
The dashboard in the Home screen displays a high-level view of testing progress for
each controller as shown in Figure 62. When you click on any of the controllers,
further test details will be displayed.
Page| 8181
Figure 62: Dashboard
Page 89
82
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
Display the different types of reports for a specific search
criterion.
View Bookmarks
In the Home screen, you can also see the Bookmarks pane (Figure 63) that will
display previously marked test results for quick reference. The Bookmarks pane
contains complete test details and controller information used during the relevant
test.
Figure 63: Bookmarks Pane
When you click on a test that is bookmarked, you will see the test details.
To bookmark a test, click when the test is being executed. For details
on how to execute tests, refer to the Performance Tests chapter.
The following table displays the types of reports generated and other features of the
reporting module.
The Robustness Summary Report screen shows the summary information for the
modes in which the robustness tests have been executed.
View Individual Test Execution Report
The individual test report displays the test results pertaining to a specific test
execution. The tool captures the results of all test executions and these results are
stored for analysis.
The reports generated by the PktBlaster SDN Controller Test pertain to the test
executions performed in the last one week. However, you can change the period for
which the report is generated by:
Selecting one of the pre-defined time periods
Specifying a date range
To specify a time period or a date range to generate reports:
1. Click the default date range displayed above the Execution History pane as
indicated in the Reports screen.
The Search Options screen is now displayed.
Figure 76: Reports Screen
Page 99
92
PktBlaster SDN Controller Test-Performance Benchmarking User Guide
You can download a report that is displayed in PDF format.
To download a report, click at the top of the relevant report as shown
in the Download Report screen.
Figure 79: Download Report
The report is now downloaded to your system. The PDF Report Figure 80 displays
details of Controller Information, Test Configuration and participating Switch Group
Network(s) along with Performance Chart.
Loading...
+ hidden pages
You need points to download manuals.
1 point = 1 manual.
You can buy points or you can get point for every manual you upload.