|August 30, 2017||Latency||OCI Lab 2017||3.12||lab170830|
|March 17, 2010||Latency
loopback: host1 == host2
|SGI Lab, 3.3 GHz||2.1.2||sgi100317|
|January 27, 2010||Latency
loopback: host1 == host2
|SGI Lab, 3.0 GHz||2.0||sgi100127|
|January 25, 2010||Latency||OCI Lab||2.0||lab100125|
Testing using the OpenDDS-Bench performance testing framework is based on the use of configuration files to specify the behavior and topology of the test. There are two basic types of configuration file used for the testing. These are transport configuration files, used to configure the instances of transports for the test processes, and test configuration files, used to establish the test conditions and topology. These are briefly described here.
Transport Configuration Files
The transport configuration files specify the specific configuration of transport instances by instance key. Each transport used within a test will have a transport key associated with it, and that key is configured with a section in a transport configuration file. Typically a single transport configuration file is used per test, with each process selecting the appropriately configured transport instances that it uses from the file. The following table shows the transport configuration files provided in the etc directory for use during testing.
|general purpose configuration file defining 10 transport instances of each type|
|configures transport instances using the tcp transport type|
|configures transport instances using the
due to the nature of this transport implementation, this file will need to be tailored for each test host on which it is used
|configures transport instances using the multicast transport type to operate in best effort mode|
|configures transport instances using the multicast transport type to operate in reliable mode|
Test Configuration FilesThe test configuration files used for each test are unique are are defined as part of the OpenDDS-Bench performance test framework. Each test will have a configuration file to be used for each test process for each test case. The configuration files specify both the DDS Quality of Service policy values for each Entity in the test process and the test specific configuration for that process and those Entities. The data which can be specified in these configuration files is defined in the $DDS_ROOT/performance-tests/Bench/doc/datamodel.txt file.
Performance tests of the OpenDDS implementation were executed in various environments. The environments for which we are reporting results are described in the following table.
|OCI test lab||SUSE 9.1, Intel P4 2.6 GHz, 1.5 GB RAM, gigabit ethernet|
|OCI test lab 2017||CentOS 7, Intel Core2 Duo E8400 3.00 GHz, 5.5 GB RAM, gigabit ethernet|
|SGI 3.0 GHz test lab||SUSE SLES 10, SGI Altix ICE 8200, Intel Harpertown (Xeon 3.0 GHz), 2 GB RAM|
|SGI 3.3 GHz test lab||SUSE SLES 10, SGI Altix ICE 8200, Intel W5590 (Xeon 3.3 GHz), 2 GB RAM|
The tests available with configurations in the OpenDDS-Bench performance test framework are summarized in the table below.
|Latency||Measure latency between 2 hosts from start of send to end of read.|
|Latency (loopback)||Measure latency using loopback interface from start of send to end of read. This is the same test as the 2 host test, but with host1 == host2.|