pings faster then 1/s?

Trs80

unread,

Dec 15, 2008, 11:36:04 PM12/15/08

to

I want to measure the latency on a LAN as traffic increases. Is there a way
to make a continuous ping occur faster then once/second? Or is there
another instruction that I can use to send a short message from one NIC to
another on a LAN with a settable frequency of occurrence? My plan is to use
a sniffer, WireShark to measure the round trip time between each ping and
the response.

Any tips appreciated.
thanks!

Ted Davis

unread,

Dec 16, 2008, 9:16:37 AM12/16/08

to

Are you sure you know what "latency" means in the context of a LAN? A
single ping (or the average of a few pings) when the LAN is quiet provides
the basic latency between the pinging location and the target, where
"latency" is defined as the interval between a stimulus and detection of
its response. As load increases, latency *is* increased somewhat, but most
of the LAN slowdown is due to other factors, including packet loss,
routing delays, and server loading - ping is not affected by software
slowdowns: it sees mostly basic latency + routing delays (due to load),
and packet losses (also load sensitive). It is quite possible for ping to
show little deteriation in latency while the LAN itself is barely
functional because of load concentration on points that can respond to
packets at the NIC level at full network speed, but are functionally
overloaded at the software level.

Increasing the ping rate will have little effet on the numbers it produces
except that the packets contribute to media loading. This is one of the
reasons ping has a default rate of once per second and a method of
extending the interval, but not decreasing it.

Since LAN performance is widely needed information, most routers can
supply the statistics in a readily usable form for intersegment traffic
just as smart switches provide it for intrasegment traffic.

If you aren't using smart switches and routers, your LAN is either too
small for statistics to mean much or it is so badly equiped that the
infrastructure itself is likely the main problem. If you have traffic
congestion problems with older hardware - slowdowns that are *not*
attributable to inadequate servers - then an upgrade to 100baseT (or
better) with smart switches and routers will almost certainly fix your
prohlems. If you do have smart switches and routers, then their
statistics will help diagnose the problem.

--
T.E.D. (tda...@mst.edu)

Trs80

unread,

Dec 16, 2008, 10:22:43 AM12/16/08

to

yes I know exactly what it means. Is there a way to create pings faster
then 1s?
"Ted Davis" <tda...@mst.edu> wrote in message
news:pan.2008.12.16....@mst.edu...

Ted Davis

unread,

Dec 16, 2008, 4:35:44 PM12/16/08

to

On Tue, 16 Dec 2008 07:22:43 -0800, Trs80 wrote:

> yes I know exactly what it means. Is there a way to create pings faster
> then 1s?

Fping (<http://www.kwakkelflap.com/fping.html>)

Not tested or verified by me.

--
T.E.D. (tda...@mst.edu)

Zaphod Beeblebrox

unread,

Dec 17, 2008, 9:01:18 AM12/17/08

to

"Trs80" <tr...@yahoo.com> wrote in message
news:6JP1l.11270$5P1....@newsfe13.iad...

You could always limit the ping to a single instance with '-n 1' and
then use a 'for /L' loop to control the number of pings yourself. That
would ping as fast as possible without using an external utility.

--
Zaphod

No matter where you go, there you are!