Microsoft has it's transmit window/buffer fixed at around 10Kbytes,
but can have its receive window go upto 65Kbytes. So the 109Kbps
limitation is only if the Microsoft TCP/IP stack is sending the data
(typically when made to act as a server). If the Solaris TCP is used
as server (socket sending the data) and either a Solaris TCP or
Microsoft TCP is receiving the data we have observed throughputs close
to 65535 x 8/0.6s = 870Kbps. Solaris TCP/IP stack is able to open both
the transmit window/buffer and receive window/buffer upto 65535 bytes.
So the maximum achievable throughputs will look like the following:
Transmit_TCP_stack Maximum Throughput
Microsoft TCP: approx. 109Kbps
Solaris TCP approx. 870Kbps
FTP uses TCP, so the throughputs are same as shown above. All the TCP
measurements we did were using FTP.
Ravi
______________________________ Reply Separator _________________________________
Subject: re: TCP over GEO < 512kbps
Author: <[email protected]> at INTERNET
Date: 11/7/97 2:39 PM
Form: Reply
Text: (13 lines follow)
Aaron, Mark et al
In addition, if it is not there already, we could add a graph/table showing
max possible throughput for a variety of window sizes with some markers of
where typical TCP stacks place their windows.
E.g. I think Microsoft adopts 8kByte window => max thruput = 8192 x 8 / 0.6s
= 109 kbps.
Sun = ?
FTP = ?
What do you think?
Matt
Original text: (52 lines follow)
>From AARON@SMTPGATE (Aaron Falk) {[email protected]}, on 6/11/97 1:13 PM:
To: MALLMAN@SMTPGATE (Mark Allman) {[email protected]}
Cc: TCP-OVER@SMTPGATE (tcp-over-satellite)
{[email protected]}
Mark-
I just returned from MILCOM where some statements were made
at several panels I attended to the effect that TCP
throughput over a GEO is limited to under 512kbps and "this
is a fact of life that we are just going to have to live
with." This is perpetuating the misinformation that first
appeared in the infamous LA Times article. Even
misinformation, if repeated often enough, can become
de-facto reality. One of the goals of this group was provide
a forum and solicit the IETF communities assistance in order
to clarify the issue.
To that end:
Your draft lays out very nicely what you'd like to have in a
TCP that runs over satellite. I can think of two tables that
might help clarify the issues a little further for non-TCP
experts:
1) a table which lists the options recommended and whether
it needs to be implemented in the server or the client and
2) a table which identifies which recommended options are
currently available in commercial stacks with a comment
section containing any additional information about stack
implementor's plans to add them. (This might be better as an
appendix since it will certainly be dated very quickly. But
even a dated snapshot is better than pure hearsay.)
Of particular interest (in debunking the <512kbps issue) is
the status of TCP large windows. Being able to say when and
where this will be available will support the argument that
a special "satellite" TCP isn't necessary since these
changes are already being put into stacks to support gigabit
terrestrial nets.
aaron
-- Aaron Falk (310) 814-4932 TRW, Inc Electronics Systems & Technology Division [email protected] Use Proportional Font: true Previous From: HALSEM @ INTELSAT (matthew halsey) Previous To: [email protected],[email protected] Original to: MALLMAN@SMTPGATE (Mark Allman) {[email protected]} Previous Cc: [email protected] Original cc: TCP-OVER@SMTPGATE (tcp-over-satellite) {[email protected]} Attachment Count: 0
Received: from neal.ctd.comsat.com (134.133.40.21) by smtpgw.cws.comsat.com with SMTP (IMA Internet Exchange 2.11 Enterprise) id 00088D12; Fri, 7 Nov 97 16:50:20 -0500 Received: from achtung.sp.trw.com [129.4.51.2] by neal.ctd.comsat.com with smtp (Exim 1.71 #2) id 0xTwIA-0002pe-00; Fri, 7 Nov 1997 16:50:14 -0500 Received: by achtung.sp.trw.com (4.1/SMI-4.1) id AA10906; Fri, 7 Nov 97 11:36:25 PST Received: from intelsat1.intelsat.int by achtung.sp.trw.com (4.1/SMI-4.1) id AA10901; Fri, 7 Nov 97 11:36:21 PST Received: (from smap@localhost) by intelsat1.intelsat.int (8.6.10/8.6.10) id OAA18843; Fri, 7 Nov 1997 14:38:20 -0500 Received: from smtpgate.adm.intelsat.int(164.86.14.240) by intelsat1 via smap (V1.3mjr) id sma007322; Fri Nov 7 14:37:42 1997 Received: from Connect2 Message Router by smtpgate.adm.intelsat.int via Connect2-SMTP 4.30A; Fri, 7 Nov 1997 14:40:15 -0500 Message-Id: <[email protected]> In-Reply-To: <[email protected]> Date: Fri, 7 Nov 1997 14:39:00 -0500 From: matthew halsey <[email protected]> Reply-To: <[email protected]> Organization: INTELSAT To: [email protected], [email protected] Cc: [email protected] Subject: re: TCP over GEO < 512kbps Importance: Normal Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="mime-boundary-iaanaabhmb-00EA2CA4" X-Mailer: Connect2-SMTP 4.30A MHS/SMF to SMTP Gateway Sender: [email protected] Precedence: bulk
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:31 EST