52.23.22 UDP

Data transfer testing using the UDP protocol is a good way to determine the optimum data throughput rate for services that tolerate a given degree of data loss. UDP is more suitable than TCP for this purpose, since the rigorous acknowledgement and retransmission mechanisms of TCP (designed to guarantee full data integrity) will always slow down data throughput, even if the service does not require these safeguards.
You can let TEMS Investigation run the UDP test in an automated fashion. The application will then start out at a low throughput rate and gradually increase the throughput until the packet loss becomes excessive.
UDP testing is done against a PC that is running the application TEMS UDP Server. How to install and configure this software is described in the document "TEMS UDP Server User's Manual", found in the TEMS Investigation documentation package. The TEMS Investigation application has a UDP client built into it which interacts with this server.

Configuration property:

Remote Address: The IP address or host name of the UDP server.

TCP Control Port: The TCP protocol is used in setting up the UDP session and also to communicate the test results after the session has concluded. Specify the TCP control port to use here.

UDP Mode:

Automatic: The TEMS Investigation UDP client will automatically determine the optimum data rate; see the introductory text above. In this case, the parameters from Packets Per Second onward (see below) are set automatically and are not editable. Please note that you cannot test Full Duplex in automatic mode.

Manual: You specify the data rate and packet rate yourself, using the parameters below.

Direction:

Send: The client sends UDP data to the server.

Receive: The server sends UDP data to the client.

Full Duplex: Data is sent in both directions concurrently.

Packet Size: The UDP packet size (in bytes) that will be used.

Packets per Second: Number of UDP packets to send or receive per second. For Full Duplex mode, this is the packet rate in each direction.

Throughput (kbit/s): The UDP throughput rate that will be used.

The packet and throughput parameters work as follows. If you select automatic mode, all parameters are set by the application and cannot be edited. If you select manual mode, you specify the parameter values yourself. However, since the parameters are constrained by the relation (packets per second) ´ (packet size) = throughput, when you change one parameter the others will be adjusted so that the above equation still holds true.

Duration: The duration of the UDP session ("hh:mm:ss").
Lists of UDP information elements: download/upload.
List of UDP events.
Further Notes on UDP Testing
Statistics on completed UDP sessions are output on the Summary tab of the Service Control Monitor, just as for other service sessions. See section "Viewing Statistics on Script Execution".
It must be kept in mind that the uplink UDP information elements in the category Data, which are updated continuously in the course of the UDP session (see Information Elements and Events, section "Information Element Table") only indicate the amount of data sent. They do not indicate the amount successfully transferred to the server. Therefore, when the packet loss rate becomes non-negligible, the "UDP Upload Throughput" information element is no longer reliable. To find out what percentage of packets was actually received by the UDP server, you need to check the summary statistics in the Service Control Monitor. The client uses TCP to obtain this information from the server after the UDP session has concluded.
By contrast, the "UDP Download Throughput" is always accurate.

Note:UDP upload testing tends to cause very high CPU load, especially when multiple users are running UDP concurrently.