5. Command-Line Parameters

Several parameters are available; you can enter them on the command line. All parameters are optional and case-insensitive; partial matches are allowed (such as -h instead of -help).

Some parameters only make sense in the publishing or subscribing application. The parameters are presented in the following tables, based on whether they may be used in a publishing application, a subscribing application, or both:

• Test Parameters for Publishing and Subscribing Applications
• Test Parameters Only For Publishing Applications
• Test Parameters Only For Subscribing Applications
• Test Parameters to Control RTI Connext DDS Secure Options

As you will see in the tables, the -pub parameter specifies a publishing application and the -sub specifies a subscribing application. If you do not specify See -pub then -sub is assumed.

For additional information on setting the parameters, see sections:

• Publication rate and Spinning vs. Sleeping
• Send-Queue Size and Queue-Full Behavior
• Number of Iterations vs. Latency Count
• Warming Up
• WaitSet Event Count and Delay
• How to Measure Latency for a Given Throughput
• Auto Tuning and Turbo Mode
• Optimizing Your OS For Network Performance

5.1. Test Parameters for Publishing and Subscribing Applications

• -bestEffort

  Use best-effort communication.

  Default: false (use reliable communication).

  For an introduction to the RTI reliability model, see the Strict Reliability design pattern in the RTI Connext DDS Core Libraries Getting Started Guide. See also: Reliable Communications in the RTI Connext DDS Core Libraries User’s Manual.

• -dataLen <bytes>

  Length of payload in bytes for each send.

  Default: 100 bytes.

  Range: 28 - 2147483128 bytes

  The lower limit is the number of “overhead” bytes in the message (i.e., the timestamp, sequence number, and other meta-data used by the test); the upper limit ensures that, when the overhead of the wire protocol is added, it doesn’t overflow the UDP maximum datagram size of 64KB.

  If <bytes> is bigger than 63000 RTI Perftest will enable the use of Asynchronous Publishing and Unbounded Sequences. If See -scan is specified, this value is ignored.

• -verbosity

  Run with different levels of verbosity for RTI Connext DDS.

  0 - SILENT

  1 - ERROR (default)

  2 - WARNING

  3 - ALL

• -dynamicData

  Run using the Dynamic Data API functions instead of the rtiddsgen generated calls.

  Default: false

• -durability <0|1|2|3>

  Sets the Durability kind:

  0 - VOLATILE (default)

  1 - TRANSIENT LOCAL

  2 - TRANSIENT

  3 - PERSISTENT

  For an introduction to the RTI durability model, see the Historical Data design pattern in the RTI Connext DDS Core Libraries Getting Started Guide. See also: Mechanisms for Achieving Information Durability and Persistence, Chapter 12, in the RTI Connext DDS Core Libraries User’s Manual.

• -domain <ID>

  Domain ID.

  The publisher and subscriber applications must use the same domain ID in order to communicate.

  Default: 1

  Range: 0 - 200

  See Choosing a Domain ID and Creating Multiple Domains, Section 8.3.4, in the RTI Connext DDS Core Libraries User’s Manual.

• -help

  Print an informative message with all the available command-line parameters and exit.

• -instances <int>

  Set the number of instances to use in the test. The publishing and subscribing applications must specify the same number of instances.

  This option only makes sense when testing a keyed data type; to do so, use See -keyed.

  Default: 1

  Range: > 0

• -keyed

  Specify the use of a keyed type.

  Default: Unkeyed type.

• -multicast

  Use multicast to receive data. In addition, the Datawriter heartbeats will be sent using multicast instead of unicast.

  The following default multicast addresses will be used for each of the topics:

  | **latency:** ``239.255.1.2``
  | **throughput:** ``239.255.1.1``
  | **announcement:** ``239.255.1.100``
  

  See -multicastAddr <address> for how to change these IP addresses.

  Default: Do not use multicast.

• -multicastAddr <address>

  Enable the use of multicast. In addition, the Datawriter heartbeats will be sent using multicast instead of unicast.

  The <address> will be used by the 3 topics latency:, throughput: and announcement:.

  Default: Do not use multicast.

• -noDirectCommunication

  Indicates if the subscribing application will receive samples from the publishing application when RTI Persistence Service is used.

  Only applies when -durability <0|1|2|3> is TRANSIENT (2) or PERSISTENT (3).

  If set to true (the default), the subscribing application gets samples from the publishing application and RTI Persistence Service. This mode provides low latency between endpoints.

  If set to false, the subscribing application only gets samples from RTI Persistence Service. This brokered communication pattern provides a way to guarantee eventual consistency.

  Default: true (direct communication)

• -noPositiveAcks

  Disable use of positive ACKs in the reliable protocol.

  Default: true (use positive ACKs)

  See -qosFile <filename> option for more information.

• -noPrintIntervals

  Prevent printing of statistics at intervals during the test.

  By default, statistics are printed every second in the subscribing application, and after receiving every latency echo in the publishing application.

• -qosFile <filename>

  Path to the XML file containing DDS QoS profiles.

  Default: perftest_qos_profiles.xml

  The default file contains these QoS profiles:

  The ThroughputQos, LatencyQos, and AnnouncementQos profiles are used by default.

  Note: some QoS values are ‘hard-coded’ in the application, therefore setting them in the XML file has no effect; see the See Note:.

  See comments in perftest_qos_profiles.xml, as well as Configuring QoS with XML, Chapter 17 in the RTI Connext DDS Core Libraries User’s Manual.

• qosLibrary <library name>

  Name of QoS Library for DDS Qos profiles

  Default: PerftestQosLibrary

• -noXmlQos

  Avoid loading the QoS from the xml profile, instead, they will be loaded from a string in code.

  This implies that changes in the XML profile will not be used.

  This option is recommended for OS without a file-system.

  Noste: This option is only present in C++ traditional and new PSM.

• -useReadThread

  Use a separate thread (instead of a callback) to read data.

  See WaitSet Event Count and Delay.

  Default: use callback for subscriber

• -waitsetDelayUsec <usec>

  Process incoming data in groups, based on time, rather than individually.

  Only used if the See -useReadThread option is specified on the subscriber side.

  See WaitSet Event Count and Delay.

  Default: 100

  Range: >= 0

• -waitsetEventCount <count>

  Process incoming data in groups, based on the number of samples, rather than individually.

  Only used if the See -useReadThread option is specified on the subscriber side.

  See WaitSet Event Count and Delay.

  Default: 5

  Range: >= 1

• -asynchronous

  Enable asynchronous publishing in the DataWriter QoS.

  Default: Not set

• -flowController <flow>

  Specify the name of the flow controller that will be used by the DataWriters. This will only have effect if the DataWriter uses Asynchronous Publishing either because it is using samples greater than 63000 Bytes or because the -asynchronous option is present.

  There are several flow controllers predefined:

  [‘default’,‘10Gbps’,‘1Gbps’].

  Default: default

  Values: ['default','10Gbps','1Gbps']

• -cpu

  Display the cpu used by the RTI Perftest process.

  Default: not set

• -unbounded <allocation_threshold>

  Use Unbounded Sequences in the data type of IDL..

  Default: 2*dataLen up to 63000 bytes. Range: 28 - 63000 bytes

• -peer <address>

  Adds a peer to the peer host address list. This argument may be repeated to indicate multiple peers.

  Default: Not set. RTI Perftest will use the default initial peers (localhost, shared-memory and multicast).

5.1.1. Transport Specific Options

By default, RTI Perftest will try to use the transport settings provided via the xml configuration file. However, it is possible to override these values directly by using the Transport spececific command-line parameters.

• -transport <TRANSPORT NAME>

  Set the transport to be used. The rest of the transports will be disabled.

  Options: UDPv4, UDPv6, SHMEM, TCP, TLS, DTLS and WAN.

  Default: Transport defined in the XML profile. (UDPv4 and SHMEM if no changes).

• -nic <ipaddr>

Restrict RTI Connext DDS to sending output through this interface. The value should be the IP address assigned to any of the available network interfaces on the machine. On UNIX systems the name of the interface is also valid. This command line parameter is mapped to the “allow_interfaces_list” property in RTI Connext DDS.

By default, RTI Connext DDS will attempt to contact all possible subscribing nodes on all available network interfaces. Even on a multi-NIC machine, the performance over one NIC vs. another may be different (e.g., Gbit vs. 100 Mbit), so choosing the correct NIC is critical for a proper test.

• -transportVerbosity <level>

Especific verbosity of the transport plugin.

Default: 0 (Errors only).

• -transportServerBindPort <port>

  For TCP and TLS. Port used by the transport to accept TCP/TLS connections.

  Default: 7400

• -transportWan

  For TCP and TLS. Use tcp across LANs and firewalls.

  Default: Not set, LAN Mode.

• -transportPublicAddress <ipaddr>

  For TCP and TLS. Public IP address and port (WAN address and port) (separated by ‘:’) associated with the transport instantiation.

  Default: Not set

• -transportWanServerAddress <ipaddr>

  For WAN transport. Address where to find the WAN Server.

  Default: Not set

• -transportWanServerPort <ipaddr>

  For WAN transport. Port where to find the WAN Server.

  Default: Not set

• -transportWanId <id>

  For WAN transport. Id to be used for the WAN transport. Required when using WAN.

  Default: Not set

• -transportSecureWan

  For WAN transport. Use DTLS security over WAN.

  Default: Not set

• -transporCertAuthority <file>

  For TLS, DTLS and Secure WAN. Certificate authority file to be used by TLS.

  Default for Publisher: ./resource/secure/pub.pem

  Default for Subscriber: ./resource/secure/sub.pem

• -transporCertFile <file>

  For TLS, DTLS and Secure WAN. Certificate file to be used by TLS.

  Default: ./resource/secure/cacert.pem

• -transporPrivateKey <file>

  For TLS, DTLS and Secure WAN. Private key file to be used by TLS.

  Default for Publisher: ./resource/secure/pubkey.pem

  Default for Subscriber: ./resource/secure/subkey.pem

5.2. Test Parameters Only For Publishing Applications

• -batchSize <bytes>

  Enable batching and set the maximum batched message size. Disabled automatically if using large data.

  Default: 0 (batching disabled)

  Range: 1 to 63000

  For more information on batching data for high throughput, see the High Throughput design pattern in the RTI Connext DDS Core Libraries Getting Started Guide. See also: How to Measure Latency for a Given Throughput and the BATCH QosPolicy, Section 6.5.2 in the RTI Connext DDS Core Libraries Getting User’s Manual.

• -enableAutoThrottle

  Enable the Auto Throttling feature. See Auto Tuning and Turbo Mode.

  Default: feature is disabled.

• -enableTurboMode

  Enables the Turbo Mode feature. See Auto Tuning and Turbo Mode. When turbo mode is enabled, See -batchSize <bytes> is ignored. Disabled automatically if using large data or asynchronous.

  Default: feature is disabled.

• -executionTime <sec>

  Allows you to limit the test duration by specifying the number of seconds to run the test.

  The first condition triggered will finish the test: -numIter or -executionTime <sec>.

  Default: 0 (i.e. don’t set execution time)

• -latencyCount <count>

  Number samples to send before a latency ping packet is sent.

  See Number of Iterations vs. Latency Count.

  Default: -1 (if -latencyTest is not specified, automatically adjusted to 10000 or -numIter whichever is less; if -latency Test is specified, automatically adjusted to 1).

  Range: must be <= -numIter

• -latencyTest

  Run a latency test consisting of a ping-pong.

  The publisher sends a ping, then blocks until it receives a pong from the subscriber.

  Can only be used on a publisher whose pidMultiPubTest = 0 (see See -pidMultiPubTest <id>).

  Default: false

• -numIter <count>

  Number of samples to send.

  See Number of Iterations vs. Latency Count and See Warming Up.

  If you set scan = true, you cannot set this option (See -scan).

  Default: 100000000 for throughput tests or 10000000 for latency tests (when -latencyTest is specified); also, see -executionTime

  Range: latencyCount (adjusted value) or higher (see -latencyCount <count>).

• -numSubscribers <count>

  Have the publishing application wait for this number of subscribing applications to start.

  Default: 1

• -pidMultiPubTest <id>

  Set the ID of the publisher in a multi-publisher test.

  Use a unique value for each publisher running on the same host that uses the same domain ID.

  Default: 0

  Range: 0 to n-1, inclusive, where n is the number of publishers in a multi-publisher test.

• -pub

  Set test to be a publisher.

  Default: -sub

• -pubRate <sample/s>:<method>

  Limit the throughput to the specified number of samples per second. The method to control the throughput rate can be: ‘spin’ or ‘sleep’.

  If the method selected is ‘sleep’, RTI Perftest will control the rate by calling the sleep() function between writing samples. If the method selected is ‘spin’, RTI Perftest will control the rate by calling the spin() function (active wait) between writing samples.

  Note: The resolution provided by using ‘spin’ is generally better than the ‘sleep’ one, specially for fast sending rates (where the time needed to spend between sending samples is very small). However this will also result in a higher CPU consumption.

  Default samples: 0 (no limit)

  Range samples: 1 to 10000000
  Default method: spin

  Values method: spin or sleep

• -scan <size1>:<size2>:...:<sizeN>

  Run test in scan mode. The list of sizes is optional and can be either in the [32,63000] range or the [63001,2147483128] range (Large Data cannot be tested in the same scan test as small data sizes). Default values to test with are ‘32:64:128:256:512:1024:2048:4096:8192:16384:32768:63000’ The -executionTime parameter is applied for every size of the scan. If -executionTime is not set, a timeout of 60 seconds will be applied.

  Default: false (no scan)

• -sendQueueSize <number>

  Size of the send queue.

  When -batchSize <bytes> is used, the size is the number of batches.

  See Send-Queue Size and Queue-Full Behavior.

  Default: 50

  Range: [1-100 million] or -1 (indicating an unlimited length).

• -sleep <millisec>

  Time to sleep between each send.

  See Spinning vs. Sleeping.

  Default: 0

  Range: 0 or higher

• -writerStats

  Enable extra messages showing the Pulled Sample Count of the Writer in the Publisher side.

  The frequency of these log messages will be determined by the -latencyCount since the message is only shown after a latency ping.

  Default: Not enabled

• -writeInstance <instance>

  Set the instance number to be sent.

  Default: Round-Robin schedule

  Range: 0 and instances

5.3. Test Parameters Only For Subscribing Applications

• -numPublishers <count>

  The subscribing application will wait for this number of publishing applications to start.

  Default: 1

• -sidMultiSubTest <id>

  ID of the subscriber in a multi-subscriber test.

  Use a unique value for each subscriber running on the same host that uses the same domain ID.

  Default: 0

  Range: 0 to n-1, inclusive, where n is the number of subscribers in a multi-subscriber test.

• -sub

  Set test to be a subscriber.

  Default: -sub

• -cft <start>:<end>

  Use a Content Filtered Topic for the Throughput topic in the subscriber side Specify 2 parameters: and to receive samples with a key in that range. Specify only 1 parameter to receive samples with that exact key.

  Default: Not set

5.4. Test Parameters to Control RTI Connext DDS Secure Options

• -secureEncryptDiscovery

  Encrypt discovery traffic.

  Default: Not set.

• -secureSign

  Sign discovery and user data packages.

  Default: Not set.

• -secureEncryptData

  Encrypt at the user data level.

  Default: Not set.

• -secureEncryptSM

  Encrypt at the RTPS sub-message level.

  Default: Not set.

• -secureGovernanceFile <file>

  Governance file. If specified, the authentication, signing, and encryption arguments are ignored. The governance document configuration will be used instead.

  Default: Not set.

• -securePermissionsFile <file>

  Permissions file to be used.

  Default for Publisher: ./resource/secure/signed_PerftestPermissionsPub.xml

  Default for Subscriber: ./resource/secure/signed_PerftestPermissionsSub.xml

• -secureCertAuthority <file>

  Certificate authority file to be used.

  Default for Publisher: ./resource/secure/pub.pem

  Default for Subscriber: ./resource/secure/sub.pem

• -secureCertFile <file>

  Certificate file to be used.

  Default: ./resource/secure/cacert.pem

• -securePrivateKey <file>

  Private key file to be used.

  Default for Publisher: ./resource/secure/pubkey.pem Default for Subscriber: ./resource/secure/subkey.pem

5.5. Additional information about the parameters

5.5.1. Secure Certificates, Governance and Permission Files

RTI Perftest provides a set of already generated certificates, governance and permission files to be loaded when using the RTI Connext DDS Secure Libraries. Both governance files and permission files are already signed, so no action is required by the user. These files are located in $(RTIPERFTESTHOME)/resource/secure.

In addition to the already signed governance and permission files, the original files are also provided (not signed) as well as a bash script with the steps to generate all the signed files. Those files can be found in $(RTIPERFTESTHOME)/resource/secure/input; the script is in $(RTIPERFTESTHOME)/resource/secure/make.sh.

5.5.2. Publication rate and Spinning vs. Sleeping

When the publisher is writing as fast as it can, sooner or later, it is likely to get ahead of the subscriber. There are 4 things you can do in this case:

1. Nothing – for reliable communication, write() will block until the subscriber(s) catch up.
2. Slow the writing down by sleeping (See -sleep <millisec>). This approach is friendlier to the other processes on the host because it does not monopolize the CPU. However, context switching is expensive enough that you can’t actually “sleep” for amounts of time on the order of microseconds, so you could end up sleeping too long and hurting performance. (Operating systems (including Linux and Windows) have a minimum resolution for sleeping; i.e., you can only sleep for a period of 1 or 10 ms. If you specify a sleep period that is less than that minimum, the OS may sleep for its minimum resolution.)
3. Set a publication rate (See -pubRate <count>:<method>). This approach will make RTI Perftest automatically set the rate of the write call so you can get the number of samples per second requested (if possible). This option allows to choose to use sleep() between calls or spin(). This second approach will add a pause without yielding the CPU to other processes, making it easier to “sleep” for very short periods of time. Avoid spinning on a single-core machine, as the code that would break you out of the spin may not be able to execute in a timely manner.
4. Let the publisher automatically adjust the writing rate (See -enableAutoThrottle). This option enables the Auto Throttle feature introduced in RTI Connext DDS 5.1.0 and its usage is preferred over See -spin <count> because the amount of spin is automatically determined by the publisher based on the number of unacknowledged samples in the send queue.

See also: Send-Queue Size and Queue-Full Behavior.

5.5.3. Send-Queue Size and Queue-Full Behavior

In many distributed systems, a data producer will often outperform data consumers. That means that, if the communications are to be reliable, the producer must be throttled in some way to allow the consumers to keep up. In some situations, this may not be a problem, because data may simply not be ready for publication at a rate sufficient to overwhelm the subscribers. If you’re not so lucky, your publisher’s queue of unacknowledged data will eventually fill up. When that happens, if data is not to be lost, the publication will have to block until space becomes available. Blocking can cost you in terms of latency.

To avoid the cost of blocking, consider the following:

• Enlarge your publisher’s queue (See -sendQueueSize <number>). Doing so will mean your publisher has to block less often. However, it may also let the publisher get even further ahead of slower subscribers, increasing the number of dropped and resent packets, hurting throughput. Experimenting with the send queue size is one of the easy things you can do to squeeze a little more throughput from your system.
• Enable Auto Throttling (See -enableAutoThrottle). This option enables the Auto Throttle feature introduced in RTI Connext DDS 5.1.0. When this option is used, the publisher automatically adjusts the writing rate based on the number of unacknowledged samples in the send queue to avoid blocking.

Note:

The following values in the DataWriterProtocolQosPolicy are ‘hard-coded’ in the application, therefore setting these values in the XML QoS profile will have no effect:

• rtps_reliable_writer.heartbeats_per_max_samples is set to (sendQueueSize/10)
• rtps_reliable_writer.low_watermark is set to (sendQueueSize * 0.10)
• rtps_reliable_writer.high_watermark is set to (sendQueueSize * 0.90)

For more information on the send queue size, see the RESOURCE_LIMITS QosPolicy, Section 6.5.20 in the RTI Connext DDS Core Libraries User’s Manual (specifically, the max_samples field).

5.5.4. Number of Iterations vs. Latency Count

When configuring the total number of samples to send during the test (See -numIter <count>) and the number of samples to send between latency pings (See -latencyCount <count>), keep these things in mind:

• Don’t send latency pings too often. One of the purposes of the test is to measure the throughput that the middleware is able to achieve. Although the total throughput is technically the total data sent on both the throughput and latency topics, for the sake of simplicity, the test measures only the former. The implicit assumption is that the latter is negligible by comparison. If you violate this assumption, your throughput test results will not be meaningful.
• Keep the number of iterations large enough to send many latency pings over the course of the test run. Your latency measurements, and the spread between them, will be of higher quality if you are able to measure more data points.
• When selecting See -numIter <count>, choose a value that allows the test to run for at least a minute to get accurate results. Set See -numIter <count> to be millions for small message sizes (<1k); reduce as needed for larger sizes (otherwise the tests will take longer and longer to complete).

5.5.5. Warming Up

When running the performance test in Java, and to a lesser extent, C#, you may observe that throughput slowly increases through the first few incremental measurements and then levels off. This improvement reflects the background activity of the just-in-time (JIT) compiler and optimizer on these platforms. For the best indication of steady-state performance, be sure to run the test for a number of samples (See -numIter <count>) sufficient to smooth out this start-up artifact.

5.5.6. WaitSet Event Count and Delay

RTI Connext DDS, and by extension, this performance test, gives you the option to either process received data in the middleware’s receive thread, via a listener callback, or in a separate thread (See -useReadThread) via an object called a WaitSet. The latter approach can be beneficial in that it decouples the operation of your application from the middleware, so that your processing will not interfere with Connext DDS’s internal activities. However, it does introduce additional context switches into your data receive path. When data is arriving at a high rate, these context switches can adversely impact performance when they occur with each data sample.

To improve efficiency, the command line parameters -waitsetDelayUsec <usec> and -waitsetEventCount <count> allow you to process incoming data in groups, based on the number of samples and/or time, rather than individually, reducing the number of context switches. Experiment with these values to optimize performance for your system.

For more information, see these sections in the RTI Connext DDS Core Libraries User’s Manual: Receive Threads (Section 19.3) and Conditions and WaitSets (Section 4.6).

5.5.7. How to Measure Latency for a Given Throughput

If you want to measure the minimum latency for a given throughput, you have to use the command-line parameters -sleep <millisec>, -spin <count> and -batchSize <bytes> to experimentally set the throughput level for a given test run.

For example, suppose you want to generate a graph of latency vs. throughput for a packet size of 200 bytes and throughput rates of 1000, 10K, 20K, 50K, 100K, 500K, and Max messages per second.

For throughput rates under 1000 messages per second, use -sleep <ms> to throttle the publishing application. For example, -sleep 1 will produce a throughput of approximately 1000 messages/second; -sleep 2 will produce a throughput of approximately 500 messages/second.

For throughput rates higher than 1000 messages per second, use -spin <spin count> to cause the publishing application to busy wait between sends. The <spin count> value needed to produce a given throughput must be experimentally determined and is highly dependent on processor performance. For example -spin 19000 may produce a message rate of 10000 messages/second with a slow processor but a rate of 14000 messages/second with a faster processor.

Use batching when you want to measure latency for throughput rates higher than the maximum rates of sending individual messages. First, determine the maximum throughput rate for the data size under test without batching (omit See -batchSize <bytes>). For example, on a 1-Gigabyte network, for a data size of 200 bytes, the maximum throughput will be about 70,000 messages/sec. We will refer to this value as max_no_batch.

For all throughput rates less than max_no_batch (e.g., 70,000 messages/sec.), do not use batching, as this will increase the latency.

Use batching to test for throughput rates higher than max_no_batch: start by setting -batchSize to a multiple of the data size. For example, if the data size is 200 bytes, use -batchSize 400 (this will put 2 messages in each batch), -batchSize 800 (4 per batch), etc. This will allow you to get throughput/latency results for throughputs higher than the max_no_batch throughput rate.

Note: For larger data sizes (8000 bytes and higher), batching often does not improve throughput, at least for 1-Gigabyte networks.

5.5.8. Auto Tuning and Turbo Mode

RTI Connext DDS includes since 5.1.0 two features that allow the middleware to auto-tune the communications to achieve better performance. These features are Auto Throttling and Turbo Mode. For more information about both features, refer to Sections 10.4, Auto Throttling for DataWriter Performance – Experimental Feature and 6.5.2.4 Turbo Mode: Automatically Adjusting the Number of Bytes in a Batch – Experimental Feature in the RTI Connext DDS Core Libraries User’s Manual. The performance test application includes two command-line options to enable these features: -enableAutoThrottle and -enableTurboMode.

With Auto Throttling, the publisher automatically adjusts the writing rate based on the number of unacknowledged samples in the send queue to avoid blocking and provide the best latency/throughput tradeoff.

With Turbo Mode, the size of a batch is automatically adjusted to provide the best latency for a given write rate. For slow write rates, the batch size will be smaller to minimize the latency penalty. For high write rates, the batch size will be bigger to increase throughput. When turbo mode is used, the command line option See -batchSize <bytes> is ignored.

To achieve the best latency under maximum throughput conditions, use See -enableAutoThrottle and See -enableTurboMode in combination.

5.5.9. Optimizing Your OS For Network Performance

The network stacks of popular operating systems are not always tuned for maximum performance out of the box. RTI has found that the following configuration changes frequently improve performance for a broad set of demanding applications. Consider testing your network performance with and without these changes to learn if they can benefit your system.

5.5.9.1. Optimizing Linux Systems

Edit the file /etc/sysctl.conf and add the following:

net.core.wmem_max = 16777216
net.core.wmem_default = 131072
net.core.rmem_max = 16777216
net.core.rmem_default = 131072
net.ipv4.tcp_rmem = 4096 131072 16777216
net.ipv4.tcp_wmem = 4096 131072 16777216
net.ipv4.tcp_mem = 4096 131072 16777216

net.core.netdev_max_backlog = 30000
net.ipv4.ipfrag_high_threshold = 8388608

run /sbin/sysctl -p

5.5.9.2. Optimizing Windows Systems

1. From the Start button, select Run…, then enter regedit.
2. Change this entry: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\ Services\Tcpip\Parameters
   • Add the DWORD key: MaximumReassemblyHeaders
   • Set the value to 0xffff (this is the max value)
   • See http://support.microsoft.com/kb/811003 for more information.
3. Change this entry: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\ Services\AFD\Parameters
   • Add the DWORD key: FastSendDatagramThreshold
   • Set the value to 65536 (0x10000) See http://support.microsoft.com/kb/235257 for more information.
4. Reboot your machine for the changes to take effect.