question on tftp

Hello All!

If I know that a ping time is 1ms, then how can I calculate how much time is
needed to transfer a 500K long file with
tftp?
Thank you in advanve,
Bal

On Fri, 02 Dec 2005 21:44:37 +0100, Balage wrote:

> If I know that a ping time is 1ms, then how can I calculate how much time is
> needed to transfer a 500K long file with
> tftp?

Based on that information alone, you cannot. Latency is not a measure
of bandwidth.

On a lightly-loaded 100Mbit network you should be able to pull 500KB in
about 50-100ms.
[root@Ashtown ~]# ttcp -n 64 -st 192.168.1.3
ttcp-t: buflen=8192, nbuf=64, align=16384/0, port=5001 tcp -> 192.168.1.3
ttcp-t: socket
ttcp-t: connect
ttcp-t: 524288 bytes in 0.08 real seconds = 6779.39 KB/sec +++
ttcp-t: 64 I/O calls, msec/call = 1.21, calls/sec = 847.42
ttcp-t: 0.0user 0.0sys 0:00real 0% 0i+0d 0maxrss 0+3pf 14+0csw

On a lightly-loaded gigabit network, less than 10ms.
[root@Ashtown ~]# ttcp -n 64 -st 192.168.2.2
ttcp-t: buflen=8192, nbuf=64, align=16384/0, port=5001 tcp -> 192.168.2.2
ttcp-t: socket
ttcp-t: connect
ttcp-t: 524288 bytes in 0.01 real seconds = 100058.63 KB/sec +++
ttcp-t: 64 I/O calls, msec/call = 0.08, calls/sec = 12507.33
ttcp-t: 0.0user 0.0sys 0:00real 0% 0i+0d 0maxrss 0+3pf 14+2csw

On Fri, 02 Dec 2005 21:44:37 +0100, Balage wrote:

> If I know that a ping time is 1ms, then how can I calculate how much time is
> needed to transfer a 500K long file with
> tftp?

Based on that information alone, you cannot. Latency is not a measure
of bandwidth.

On a lightly-loaded 100Mbit network you should be able to pull 500KB in
about 50-100ms.
[root@Ashtown ~]# ttcp -n 64 -st 192.168.1.3
ttcp-t: buflen=8192, nbuf=64, align=16384/0, port=5001 tcp -> 192.168.1.3
ttcp-t: socket
ttcp-t: connect
ttcp-t: 524288 bytes in 0.08 real seconds = 6779.39 KB/sec +++
ttcp-t: 64 I/O calls, msec/call = 1.21, calls/sec = 847.42
ttcp-t: 0.0user 0.0sys 0:00real 0% 0i+0d 0maxrss 0+3pf 14+0csw

On a lightly-loaded gigabit network, less than 10ms.
[root@Ashtown ~]# ttcp -n 64 -st 192.168.2.2
ttcp-t: buflen=8192, nbuf=64, align=16384/0, port=5001 tcp -> 192.168.2.2
ttcp-t: socket
ttcp-t: connect
ttcp-t: 524288 bytes in 0.01 real seconds = 100058.63 KB/sec +++
ttcp-t: 64 I/O calls, msec/call = 0.08, calls/sec = 12507.33
ttcp-t: 0.0user 0.0sys 0:00real 0% 0i+0d 0maxrss 0+3pf 14+2csw

Balage <(E-Mail Removed)> wrote:
> If I know that a ping time is 1ms, then how can I calculate how much
> time is needed to transfer a 500K long file with tftp?

By looking at the TFTP specification (www.ietf.org) you can see how
much data TFTP will have outstanding on the network at one time and
with that you can calculate a best-case transfer time for your 500K
file.

You may also want to make sure, if at all possible, that your ping
test is run in a manner similar to that of a TFTP transfer wrt packet
sizes.

rick jones
--
Process shall set you free from the need for rational thought.
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

Balage wrote:

> If I know that a ping time is 1ms, then how can I calculate how much time
> is needed to transfer a 500K long file with
> tftp?

Insufficient data. Ping time reflects how long it takes for a short packet
to cross the wire to the other end and get a response back. It is not a
measure of bandwidth. A sluggish computer might appear to take longer to
respond to a ping than a faster one.

James Knott <(E-Mail Removed)> wrote:
> Insufficient data. Ping time reflects how long it takes for a short
> packet to cross the wire to the other end and get a response back.
> It is not a measure of bandwidth. A sluggish computer might appear
> to take longer to respond to a ping than a faster one.

While 99 times out of 10 just saying "ping time" means the time for
the ICMP Echo Reply to be recieved for a small (O(60 bytes)) ICMP Echo
Request generated by the ping utility, it could be that the ping in
question was for a larger size.

As for where the cross-over point happens for the relative importance
of network bandwidth vs latency (in either the network or the end
systems), that is a matter of experimentation.

One might reasonably ass-u-me that if the computer was sluggish
responding to an ICMP Echo Request (aka ping) it might be similarly
sluggish serving TFTP. Yes, there can be any number of altering
factors - prioritization of one traffic type over another etc...

rick jones
--
portable adj, code that compiles under more than one compiler
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...