![Metrodata NT10G05 Скачать руководство пользователя страница 55](http://html1.mh-extra.com/html/metrodata/nt10g05/nt10g05_user-manual_1779477055.webp)
NT10G05 USER MANUAL
76-02-105 Rev.B
Page 55 of 80
4.2.3.1.1.3 Training Period (L2/L3)
Whilst some services are provided as true L2 single broadcast domain networks,
most are actually provided as VPLS tunnels over MPLS core networks. When a
packet reaches the edge of such networks forwarding tables need to be configured
in order to forward the packets. Due to this connection initialisation it is possible to
see some packet loss at the beginning of a test.
However, to get around this issue a training period may be introduced before the test
is run for real. During the training period, low rate packets are transmitted to the
required loopback device without being monitored. This allows all forwarding tables
etc. to be set up such that the test will run from the first packet without loss.
The training period is defined in seconds and may be between 1 and 30 seconds.
4.2.3.1.1.4 Training Frames per Second (L2/L3)
During the training period the training FPS defines the rate of test packet generation.
It is advisable to keep this to a fairly low level, though above 1000 to ensure that no
buffers will be affected during the training period which would then be
counterproductive for the real test.
4.2.3.1.1.5 Rate (L2/L3)
The rate configuration item defines the constant rate at which packets will be
transmitted during the test. The frames are sent equally spaced, rather than in
bursts during the basic test.
The rate is defined as a percentage of the CIR defined in the SLA definition.
Note, due to minor clock differences throughout the network, different vendor
policing algorithms and background management traffic 100% CIR will almost
certainly lead to some packet loss. It is recommended that the maximum rate is set
to 97% to give some leeway and avoid packet loss.