tcp: can't explain that

Hi,

I did some time measure tests with TCP. I used the option SO_LINGER. I
did something like:

getTime
sendData
closeSocket
getTime

My Problem:
To send 80000 bytes was faster as to send 10240 bytes. Why that?
I did this not only once but nearly always it was like that:
10240 bytes : 36.5 ms
80000 bytes : 30.4 ms

Who knows why? I can't explain this to myself.

Thanks for your help.

Stephan

On Tue, 23 Sep 2003 13:33:01 +0200, Stephan Absmeier <(E-Mail Removed)>
wrote:

>My Problem:
>To send 80000 bytes was faster as to send 10240 bytes. Why that?
>I did this not only once but nearly always it was like that:
>10240 bytes : 36.5 ms
>80000 bytes : 30.4 ms
>Who knows why? I can't explain this to myself.

If the program with which you're doing this test is not too long, post
it, so that we can have a look at it, and provide some insight.


--
Fernando Gont
e-mail: (E-Mail Removed)

[To send a personal reply, please remove the ANTISPAM tag]

On Tue, 23 Sep 2003 15:30:32 -0300, Fernando Gont <(E-Mail Removed)> wrote:

>On Tue, 23 Sep 2003 13:33:01 +0200, Stephan Absmeier <(E-Mail Removed)>
>wrote:
>
>>My Problem:
>>To send 80000 bytes was faster as to send 10240 bytes. Why that?
>>I did this not only once but nearly always it was like that:
>>10240 bytes : 36.5 ms
>>80000 bytes : 30.4 ms
>>Who knows why? I can't explain this to myself.
>
>If the program with which you're doing this test is not too long, post
>it, so that we can have a look at it, and provide some insight.

The OP should also post the results of 'ifconfig'.

Transmission throughput is influenced by many things, including the local and
path MTU.


--
Lew Pitcher
IT Consultant, Enterprise Technology Solutions
Toronto Dominion Bank Financial Group

(Opinions expressed are my own, not my employers')

Hi,

Lew Pitcher schrieb:
> On Tue, 23 Sep 2003 15:30:32 -0300, Fernando Gont <(E-Mail Removed)> wrote:
>
>
>>On Tue, 23 Sep 2003 13:33:01 +0200, Stephan Absmeier <(E-Mail Removed)>
>>wrote:
>>
>>
>>>My Problem:
>>>To send 80000 bytes was faster as to send 10240 bytes. Why that?
>>>I did this not only once but nearly always it was like that:
>>>10240 bytes : 36.5 ms
>>>80000 bytes : 30.4 ms
>>>Who knows why? I can't explain this to myself.
>>
>>If the program with which you're doing this test is not too long, post
>>it, so that we can have a look at it, and provide some insight.

250 lines ;-(

tcpSender testSender2=tcpSender(argv[1],line,length);
testSender2.init(option); //socket,SO_LINGER, bind and connect
start64=getRealTime();
testSender2.work(); //send data
do
closeid=testSender2.end();//close
while(closeid<0);
ende64 = getRealTime();

this is done in a loop, the loop send 8000, 10240, 80000 and 800000
bytes. First I send 10240 bytes, close the socket, open a new one, send
the 80000 bytes
All the times it was done with the IP address, so there is no name
resolution.

>
>
> The OP should also post the results of 'ifconfig'.
ifconfig
eth0 Protokoll:Ethernet Hardware Adresse 00:50:22:AC:92:53
inet Adresse:192.168.0.111 Bcast:192.168.0.255
Maske:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:35926 errors:0 dropped:0 overruns:0 frame:0
TX packets:31672 errors:0 dropped:0 overruns:15 carrier:0
Kollisionen:0
RX bytes:16251223 (15.4 Mb) TX bytes:3055467 (2.9 Mb)

lo Protokoll:Lokale Schleife
inet Adresse:127.0.0.1 Maske:255.0.0.0
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:19081931 errors:0 dropped:0 overruns:0 frame:0
TX packets:19081931 errors:0 dropped:0 overruns:0 carrier:0
Kollisionen:0
RX bytes:778582028 (742.5 Mb) TX bytes:778582028 (742.5 Mb)


>
> Transmission throughput is influenced by many things, including the local and
> path MTU.

Ethernet 100MBit/s, sitched

>
>

Hope you can help me now.
Thanks

Stephan