Re: TCP window/stream size

From: Steve Lyons <charter21p5_at_gmail.com>
Date: Thu, 10 Sep 2009 17:34:46 -0400

RFC 1323 allows for extensions to TCP. This will allow scaling windows,
timestamps, etc. Without RFC 1323 options enabled the max you can define for
a window size is 65535.

http://www.ietf.org/rfc/rfc1323.txt

Steve Lyons

On Thu, Sep 10, 2009 at 4:07 PM, TCP IP4 <tcp.ip4_at_gmail.com> wrote:

> http://en.wikipedia.org/wiki/TCP_window_scale_option
>
>
> On Thu, Sep 10, 2009 at 12:37 PM, Christopher Copley
> <copley.chris_at_gmail.com> wrote:
> > Experts,
> >
> > I am doing some math and I am scratching my head...
> > IPv4 only provides a 16-bit (unsigned) field for the window, so you can
> only
> > increase the window size to 65535 bytes, which allows a maximum of 65535
> > data bytes to be unacknowledged in transit at any time. With a one-second
> > round-trip, that's a peak rate of 65 KB/sec, although that is 524,280
> bits
> > per second.
> > So......
> > RCV buffer size / RTT = Max TCP throughput
> > ok, now 64K times 8 bits = 512000
> > divide that by .04 (latency) and you come up with 12800000 or 12.8Mbs
> > So by using this logic any one TCP stream can never get over ~13mbs
> right?
> >
> > Is my thinking and math correct on this question?
> > So, is there any way to over come this limitation if the above is
> correct?
> >
> > Chris
> >
> >
> > Blogs and organic groups at http://www.ccie.net
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html

Blogs and organic groups at http://www.ccie.net
Received on Thu Sep 10 2009 - 17:34:46 ART

This archive was generated by hypermail 2.2.0 : Sun Oct 04 2009 - 07:42:03 ART