NDT-DEV email list created

Revision: 648
Author:   

Date:     Wed Sep 21 11:50:50 2011
Log:      Improved description of the formula used to find out when to  
temporarily stop sending packets.
http://code.google.com/p/ndt/source/detail?r=648

Modified:
 /wiki/NDTTestMethodology.wiki

=======================================
--- /wiki/NDTTestMethodology.wiki       Wed Sep 21 11:35:21 2011
+++ /wiki/NDTTestMethodology.wiki       Wed Sep 21 11:50:50 2011
@@ -50,7 +50,8 @@
     {{{
     BUFFER_SIZE * 16 < ((Next Sequence Number To Be Sent) - (Oldest  
Unacknowledged Sequence Number) - 1)
     }}}
-    The both `"Next Sequence Number To Be Sent"` and `"Oldest  
Unacknowledged Sequence Number"` values are obtained from the connection  
with the help of the [http://www.web100.org/ web100] library.
+    The both `"Next Sequence Number To Be Sent"` and `"Oldest  
Unacknowledged Sequence Number"` values are obtained from the connection  
with the help of the [http://www.web100.org/ web100] library.<br/>
+    The above mechanism is a part of the duplex mismatch detection. The  
idea for this came from Matt Mathis, and the NPAD duplex mismatch  
detection. If the server sends a maximum of 2 packets then it will never  
trigger a mismatch condition. Later when the Server-To-Client test is run a  
duplex mismatch will cause the Middlebox test to be faster than the  
Server-To-Client test.
  # After the throughput test, the server sends the following results to  
the client:
     || CurMSS || The current maximum segment size (MSS), in octets.||
     || !WinScaleSent || The value of the transmitted window scale option  
if one was sent; otherwise, a value of -1. ||
@@ -64,9 +65,7 @@

 The middlebox test's use of sequence numbers assumes that TCP Reno is  
being used.

-The formula used to find out when to temporarily stop sending packets is  
probably wrong. The idea was to use this as a part of the duplex mismatch  
detection, with a max of 2 packets in flight the Ethernet half-duplex  
sender should never see a collision so the throughput would be higher even  
though the buffer is limited.
-
-However, the formula allows for more packets in flight and it uses the  
`"Next Sequence Number To Be Sent"` instead of the `"Maximum Value of Next  
Sequence Number To Be Sent"`. The difference between these values is that  
the first one is not monotonic (and thus not a counter) because TCP  
sometimes retransmits lost data by pulling the Next Sequence Number back to  
the missing data. The latter one is the farthest forward (right most or  
largest) value of the Next Sequence Number. These values are the same  
except the situation when the Next Sequence Number is pulled back during  
recovery.
+The formula used to find out when to temporarily stop sending packets is  
probably wrong, because it allows for more packets in flight (16?).

 === Simple Firewall Test ===