Automatic Performance Tests
187
14. If you have enabled Allow multiple connections in the previous stream, then set
the Preferred window size for each TCP stream.
15. Leave the RFC 6349 configuration and Performance test panels.
16. Go to Test mode
17. Depending on your test setup (See section 7.4), configure Test me thod to One-
way (A > B), Upstream, Downstream, Bidirectional or Local bidirectional.
18. If you have configured any of the Upstream, Downstream or Bidirectional test
modes, configure the Remote unit IP.
Note: A unit configured in Remotely managed mode has to be installed and con-
figured in the remote end before any dual ended test can run (See section 7.5).
19. From the Home panel, go to CONFIG,
The port setup panel is displayed.
20. Go to Port A
21. Configure the Port A Local profile (See section 2.3).
22. If you are running a Remote bidirectional test the test configuration is ready. If you
are running a One-way (A > B) or a Local bidirectional test, then leave Port A and
go to Port B.
23. Configure the Port B Local profile (See section 2.3).
7.3.2. Setting the Performance Objectives
The TCP Throughput test based on the standard RFC 6349 provides results in terms
of three different metrics: Throughput, Efficiency and Buffer delay. All three are
described in the RFC 6349 standard:
• Throughput: Represents the maximum achievable bit rate between the client and
the server entities (or between the server and the client). Only the TCP payload
capacity is taken into account in this results. The Ethernet related fields (IFG, pre-
amble, header, and FCS), IP header and TCP header are not taken into consider-
ation. For this reason, the theoretical TCP throughput is smaller than the nominal
L1 channel capacity. For example, a 1 Gb/s Ethernet, has a maximum TCP
throughput of 949 Mb/s (1518 byte frames).
• Efficiency: It is defined as the ratio of successfully transmitted bytes to the total
number of transmitted bytes. Efficiency decreases when data has to be retrans-
mitted and the more retransmissions are necessary the smaller the TCP effi-
ciency.This parameter is closely related with frame loss events.
• Buffer delay: This parameter is the ratio between the RTT computed during the
throughput test and the inherent baseline RTT. This parameter is sensitive to wait-
ing time in intermediate buffers and congestion.
The test unit provides a Pass or a Fail result for each of these parameters based on
thresholds users can set at their will. The test unit provides one or two results for the
throughput, efficiency and buffer delay depending on the test method (See section
7.4). One of the results correspond with the upstream and the other with the down-
stream. The thresholds apply to both of them.
To Next Page