Re: 64K limit for socket send

JF Mezei <jfmezei.spamnot@xxxxxxxxxxxx> wrote:
Rick Jones wrote:
1) Is this user's _send_, plus any queued, unsent data > MSS? If yes,
send immeidately, otherwise go to question 2.

If I have a 1500 MTU (server side) and some remote dialup user has
150 MTU, isn't it better for the server to send 1500 size packets
and let the last router befofe the dialup link split the packets
versus sending 150 packets from the server ?

(Or does the TCP stack limit those packets to 150 right at the
server side ?)

The TCP stact at the dial-up user will have exchanged an MSS option of
150-40 or 110 in the SYNchronize segment and that will be used for the
connection in each direction.

At the risk of repeating what others will say, one wants to avoid IP
fragmentation when possible. All fragments of an IP datagram must
arrive at the receiver for the datagram to be reassembled. If any of
the fragments are lost, the entire datagram is toast. This
"amplifies" the packet loss rate into a rather higher datagram loss
rate, and the entire datagram must be retransmitted by something above
IP.

(sorry for using MTU here, i forget the actual number of byte
difference between MTU and MSS)

_Typically_ it will be the MTU less 20 bytes of IP header and 20 bytes
of TCP header. For IPv6 that IP header is larger, and when TCP
Timestamps are enabled on a connection, the TCP header is typically 32
bytes.

--
No need to believe in either side, or any side. There is no cause.
There's only yourself. The belief is in your own precision. - Jobert
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...
.

Relevant Pages

  • Re: SBS 2003 Setup recommendation
    ... Your suggestion about MTU looks very interesting. ... We want something lige the following setup: http://pings.dk/sbssetup.gif ... agaist SBS2003 server. ... IPSec adds a small overhead the size of a packet which would make packets ...
    (microsoft.public.windows.server.sbs)
  • Re: Problem of blocking ICMP packet while calculating Path MTU
    ... > I am in process of implementing Path MTU detection technique. ... > process the received ICMP ECHO reply packets. ... > there is no need to write server code at all. ...
    (comp.os.linux.networking)
  • Understanding path MTU discovery
    ... As far as I can tell from my reading, if a computer on the internet accesses our web server, but the reply from the server is too big (for example, the client computer is using a PPPoE link with an MTU of 1492), the client's ISP's gateway router will send an ICMP package back to our router. ... I understand that I could mark incoming packets from clients so that replies are sent out through the same interface they came in, but I would prefer to balance the output packets. ...
    (comp.os.linux.networking)
  • Re: 64K limit for socket send
    ... If I have a 1500 MTU (server side) and some remote dialup user has 150 ... isn't it better for the server to send 1500 size packets and let ... the last-hop router will be forbidden to fragment it. ...
    (comp.os.vms)
  • Re: 64K limit for socket send
    ... If I have a 1500 MTU (server side) and some remote dialup user has 150 ... isn't it better for the server to send 1500 size packets and let ...
    (comp.os.vms)