S
S
SKYnv2012-02-25 11:10:55
Data link
SKYnv, 2012-02-25 11:10:55

About the quality and performance of the Internet channel?

How can quality be tested? On different types of traffic. iperf is enough for throughput. Does anyone have a personal method? The channel is large, over several hundred megabits.

Answer the question

In order to leave comments, you need to log in

3 answer(s)
S
SKYnv, 2012-02-25
@SKYnv

mostly
http, rtmp, rtsp, ftp
to a much lesser extent
udp, smtp, pop etc. where delays, losses, etc. are not critical.
the channel is direct to the switch, which is stuck in the m9
, yes, there is a suspicion that either the iron cannot cope, or it is lousy configured, or it is cut.

B
Boris Syomov, 2012-02-25
@kotomyava

Actually iperf width, delays/losses for example mtr. The main problem in channel testing, IMHO, is to provide a point where the traffic will be consumed, in such a way that this does not affect the test result.

P
prox, 2012-02-25
@prox

The quality depends on many factors:
- physical link (optics, DSL, coax),
- equipment (model, what network adapters),
- network construction and load,
- server / client settings (different operating systems work differently).
All this must be monitored and the capabilities of the network must be known, then look for the culprit.
Tools:
Cacti - total channel load, packet types (unicast, roadcast, etc.), and losses,
Zabbix - parameter deviation monitoring,
Smokeping - latency and latency monitoring (ICMP), service operation (TCPing for WWW, probes for dns and mail,
Nuttcp/iperf - checking the available channel bandwidth and OS capabilities (they work differently: Windows XP, 7; Linux, FreeBSD, etc.)
You can also make a script for checking losses (via MTR) to find out where and when network losses appear.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question