Return-Path: 
To: 6.829-students@MIT.EDU
Subject: FAQs for Problem 1 in P2
Date: Mon, 08 Oct 2001 11:05:44 -0400
Sender: yxw@cordelia.lcs.mit.edu

Q. What are the spikes in the cwnd plot?

A. The spikes are caused by cwnd inflation. In TCP Reno's fast
recovery phase, the cwnd is inflated by returning dupacks. When TCP
Reno exists fast recovery, cwnd is reset to the right value, which is
half of the window size before the connection detects a loss. Please
ignore this transient behavior in all of your answers to this
problem. That is, this behavior is not "peculiar". 

Q. Question 1.1a asks what the RTT and bandwidth of links of Ben's
simulations are. Which RTT and links are you looking for?

A. By RTT, we mean the RTT between the TCP source node and the TCP
sink node. By bandwidth of links, it is good enough to give the
bandwidth of the bottleneck link of this connection.

Q. Which link shall I change to make the RTT 500ms?

A. It does not matter which link you change, as long as the RTT is
added up to 500ms. ( However, according to the context of the problem,
the satellite link is inside the BetaGadgets' network. I would change
the last link the connection traverses. )

Q. In figure 1, there are only two legs but in your script, there are
3 links. How should I interpret the difference?

A. Figure 1 is an abstraction that helps you to understand the queuing
behavior. It does not reflect all the details of the real network
topology. As described in 1.3, you may simply assume the latency for
each leg is 125ms and picture the second leg of the figure as the
abstraction that presents "the rest of the internet", i.e., the two
links between Ben's router and the TCP sink node. Please only consider
the packet propagation delay and the bottleneck processing/waiting
time when completing this figure.

Return-Path: 
Subject: PS2 #6 clarification
To: 6.829-students@mit.edu
Date: Thu, 11 Oct 2001 13:37:54 -0400 (EDT)

In part (f) we ask what AS do all of the routes have
in common.

What we MEAN, is "what AS shows up most frequently OTHER THAN
MIT's AS number."

So the answer to this question should NOT be MIT's AS, it should
be some other system.

Return-Path: 
Subject: More ps2#6 updates
To: 6.829-students@mit.edu
Date: Thu, 11 Oct 2001 13:53:27 -0400 (EDT)

I've updated the 'cisco_dump' file that's on the webpage;
without this update, you will not be able to answer
problem 6.3(b).  Please grab the new copy!