E
E
Erhen2014-11-18 16:58:32
Software testing
Erhen, 2014-11-18 16:58:32

Y 1564 tester from PC?

Is it possible to test Y.1564/RFC 2544 without special devices, with a normal server/host from normal PCs?
Wikipedia mentions Netround and Xena, which are very similar to ordinary downloadable programs, but something suspicious creeps in, because TCP / IP is L4, and ethernet is L2, it turns out incorrectly.

Answer the question

In order to leave comments, you need to log in

2 answer(s)
P
Pavel Kurochkin, 2014-12-18
@paulig

In principle, of course, it is possible. But you need to take into account how accurate the result will be.
The most common rfc2544 tests are Throughput, Latency, FrameLoss, Back-to-back.
All these tests are based on the so-called trial.
Each trial performs the following tasks:
And here you need to understand that hardware has an advantage in accuracy.
For example, the generation of packets from the same FPGA to the network is done with a guarantee of the inter-packet interval, that is, the accuracy of setting the load will be very high. And no suddenly launched application will affect the load.
With a delay the same situation. At the FPGA level, a timestamp is inserted at the time the packet is sent and analyzed at the time of reception. 20 nanosecond accuracy when testing from one device port to another device port is not a myth.
However, modern computers have become so powerful that they can easily create a gigabit load, and analyze it too (there may be problems at small frame sizes). Therefore, in principle, it is quite possible to drive Throughput / Frame Loss / Back-to-back with the help of a computer, if you first make sure of its performance on synthetic tests.
With 10G and 100G it will be more difficult.
But with y1564, everything is a little different. There, each trial measures all parameters at once - both received packets, and delay, and transmitted packets. Moreover, packet jitter is added to the listed ones. Here the computer is not a competitor ... In the same place, the SLA in terms of delays is at the level of about ten milliseconds, I would not trust the computer with such accuracy, and even at high load.
PS: rfc2544 and y.1564 tests are usually UDP.

V
Valentin, 2014-11-18
@vvpoloskin

And you read for yourself what testing methods are defined in the documents you mentioned. Specifically , RFC 2544 defines the following test parameters:
All these parameters can be measured with iperf, however, with a certain accuracy. It is believed that iperf gives an accuracy of 90%, especially for tcp at a large channel width.
Specialized devices provide greater accuracy, they are able to generate a lot of traffic, visually show various parameters (qos, vlan), in addition, they are certified.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question