[En-Nut-Discussion] TCP Retransmissions too fast by PC

Brett Abbott Brett.Abbott at digital-telemetry.com
Mon Apr 25 20:44:44 CEST 2005


Mark

Ethernut is throwing away the extra packet correctly.  The issue here is 
that when the Vodafone network is under strain, I end up paying for each 
and every packet that gets resent in this scenario.  This can quickly 
overload the quota.....  If I can reduce unnecessary packets from the 
PC, I save money.

Thanks
Brett

Mark E. Scott Jr. wrote:

>Ethernut isn't throwing away the resent packet when it arrives?
>
>Mark E. Scott Jr.
>mscott at awfs.net
>512-478-7727 ext. 122
>AWS, Inc.
>
>
>-----Original Message-----
>From: en-nut-discussion-bounces at egnite.de
>[mailto:en-nut-discussion-bounces at egnite.de] On Behalf Of Brett Abbott
>Sent: Saturday, April 23, 2005 4:02 AM
>To: Ethernut User Chat (English)
>Subject: [En-Nut-Discussion] TCP Retransmissions too fast by PC
>
>Hi
>
>Over a slow and performance variable network link like GPRS, we are 
>finding that the PC/Boradband Internet side of the link often gets too 
>impatient and will often send a retranmission of  a TCP packet.  At the 
>ethernut end, this often appears as a duplicate packet received shortly 
>after it has sent the ACK.  It copes ok but isnt ideal.
>
>Does anyone know of how we might better train the VB .NET application to
>
>instruct/override the TCP socket to wait longer for an ACK before 
>re-sending a packet?  Cant seem to find a way to override the Automatic 
>RTTO setting.
>
>Typical timings on the ethenut side are:
>1. Receive TCP packet
>2. Send ACK 120ms later
>3. Receive resent TCP packet 80ms after that.
>
>Propagation delay varies between 200ms and 4 seconds depending on the 
>radio sleep state.
>
>Many Thanks
>Brett
>
>  
>

-- 
-----------------------------------------------------------------
Brett Abbott, Managing Director, Digital Telemetry Limited
Email: Brett.Abbott at digital-telemetry.com
PO Box 24 036 Manners Street, Wellington, New Zealand
Phone +64 (4) 5666-860  Mobile +64 (21) 656-144
------------------- Commercial in confidence --------------------






More information about the En-Nut-Discussion mailing list