===================================================
From takhoa@rice.edu  Tue Apr 20 13:01:57 2004
===================================================

This article is basically a brief tutorial on MPLS.  For that reason,
there is not much to comment on the topic itself, i.e. MPLS.  The paper
does a good job in providing highlights of benefits of MPLS, although I
find the figures to be not very illustrative.  The issue I have with this
paper is that its focus is too broad as it tries to cover all the benefits
of MPLS.  For readers who are already familiar with MPLS, the paper is
redundant.  For readers who do not understand MPLS well, the paper glides
through some benefits of MPLS without providing enough background details
to support some of the claims.  In many sections, the paper goes into
details enough to confuse the unfamiliar readers while it fails to provide
clarifications or brief explanations of those points.  For example, the
paper mentions, but does not point out why packet overhead changes between
different technologies.  It also does not clarify whether the reduction in
overhead in MPLS is an absolute advantage over native IP routings, or
whether the overhead reduction comes at the cost of increased complexity
or other factors.

===================================================
From ahae@cs.rice.edu  Thu Apr 22 15:57:30 2004
===================================================

The article describes the Multi-protocol Label Switching
architecture (MPLS) and compares it to traditional IP
routing in several scenarios, such as DiffServ, traffic
engineering and VPN.

The author has worked on ATM networking for several years,
so it is not surprising that his arguments are mainly in
favor of MPLS. The article mostly makes high-level claims
and does not support them with any kind of experimental
evidence.

===================================================
From mittal@is.rice.edu  Fri Apr 23 15:43:13 2004
===================================================


  The article describes MPLS - Multiprotocol Label Switching. It tries to
identify the key advantages that one could obtain by using MPLS. The real
selling point of the technology according to the paper, to which I agree,
s the ability to do traffic engineering from one edge of a routing domain
to another using CR-LDP or M-RSVP. Most of the other advantages have been
offset by advances in gigabit routing, for instance.
 In my opinion MPLS is a good way of providing constraint-routing over
IP-backbones. But, it has a long way to go in terms of being accepted as a
well-known standard. It still has to answer questions like how to
effectively manage (constraint-satisfying) characteristics of LSR
(Label-switching router), and if it is to be used for QoS, it needs to
solve the end-to-end problem.