METRODATA NetTESTER FAMILY USER MANUAL
76-02-107 Rev.A
Page 48 of 133
Once the binary search has produced a result, it is normal to perform a longer test at
this rate, although this may be disabled if the fastest possible execution is required.
The longer test helps to eliminate results that are marginally higher than they should
be, produced by a combination of the short test time used in the search phase and
buffering within the network.
If the long test fails, it will be repeated, reducing the rate by 1% each time, until it
passes.
10.3.1.2.2
Latency (Delay) Test
The RFC2544 latency test runs a simple data test at a rate specified as a percentage
of the throughput test result. Average and worst-case packet delays are reported.
Note: Whilst RFC2544 calls for one packet every so often to be measured, the unit
actually measures the latency of every packet.
10.3.1.2.3
Back-to-Back (Burst Handling) Test
The back-to-back test attempts to characterise the burst-handling capability of the
link by sending single bursts of a varying number of packets. A binary search is
again employed to find the largest burst that will pass error-free.
10.3.1.2.4
Packet Sizes
Each of the three test elements is repeated for each of the selected packet sizes.
RFC2544 specifies a list of sizes (64, 128, 256, 512, 1024, 1280 and 1518 bytes),
but this can be changed, allowing for larger packets (up to 10k) or fewer sizes (to
reduce testing time), for example.
10.3.1.2.5
SLA Mode
Though RFC2544 makes no reference to the idea of a Service Level Agreement
(SLA), the unit supports an 'SLA mode' where the tests are performed in the context
of the SLA, which is used to control initial settings and assess the results of the
RFC2544 test. More specifically, the following changes occur:
•
Throughput test upper limits may be specified as a percentage of CIR, as an
alternative to an absolute value or using the media speed;
•
An 'error-free' result for a test iteration is defined as conforming to the SLA FLR,
rather than as having zero lost packets (Typically, for the shorter tests, this will
make no difference.);
•
Rates for the throughput test iterations and throughput test results are additionally
shown as a percentage of CIR in the report;
•
Details of the SLA are included in the test report;
