Re: (mobile-ip) Cache Binding..

[Dave Johnson <[e-mail]>]

Baher A. Esmat said:

>Hi all,
>	As Charles Perkins stateted in his book Mobile IP Design Principles and
>Practices, that when a mobile node's home agent intercepts a datagram and
>tunnels it to the mobile node, it may deduce that the source does not have
>a binding cache for the mobile node, so it's gonna send a binding update
>message to this source.

More to the point, the behavior of Route Optimization (including
the Binding Cache and Binding Updates) is described in the actual
specification of the protocol itself, in the Internet-Draft
draft-ietf-mobileip-optim-07.txt.  Much of the text of a number of
Internet-Drafts and RFCs written by the Mobile IP Working Group
of the IETF have been reproduced in Charlie's book, but the original
documents themselves are available by anonymous FTP and on the web.
For FTP access, go to ds.internic.net and look in the directories
"internet-drafts" or "rfc".  For web access, look under www.ietf.org.
The specific Internet-Draft documenting the behavior of the Binding
Cache and Route Optimization is available on the web at
http://www.ietf.org/internet-drafts/draft-ietf-mobileip-optim-07.txt.

>In the above caching mechanism I have a number of questions:
>
>1- Does this mean that only the first packet will be routed from the source
>to the home network then tunneled to the mobile node, whereas the rest of
>the packets will be routed directly to the mobile node as the source is
>receiving a binding cache update right after sending the 1st packet????

Typically, that is what will happen.  It is possible, however, that
the first Binding Update will be lost by the network, and so the second
packet from the source will trigger another Binding Update.  It is
also possible that the source will have sent several packets before
the Binding Update from the first arrives, causing those first few
to all be tunneled through the home agent.

>2- Does this caching technique seems to be inefficient specially if the
>number of mobile hosts grows??? If the answer is yes, would it be more
>efficient to setup a caching server to handle this stuff????

No, why do you think it might be inefficient.  Each home agent is
only involved in this for its own mobile nodes, and thus easily
scales as more home networks (and thus more mobile nodes) are added
to the Internet.  If a single home network has a very large number
of mobile nodes, it is simple to have more than one home agent
serving (each a subset of) the mobile nodes on this home network.
Each correspondent node caches only the bindings of those mobile
nodes with which it is communicating, and so this also scales well
as the number of (other) mobile nodes grows.

>3- What if the home agent was down, does the Mobile-IP propose any load
>balancing technique among a number of such hosts??? (same may applied for
>foreign agents).

I'm not sure what you mean by load balancing, but this is the same
both in the base version of Mobile IP and when using Route Optimization.
Various schemes for having backup home agents have been proposed, but
we have not carefully looked at standardizing any yet.  If the foreign
agent is down, however, it is trivial for the mobile node to simply
register with another foreign agent, if it can hear one.  It can do 
so if it simply moves to another foreign subnet, or if a subnet has
multiple foreign agents deployed, a mobile node will just automatically
register with another of them if its original foreign agent goes down.

					Dave

--
David B. Johnson                      E-mail: [e-mail]
Assistant Professor                   Home page: http://www.cs.cmu.edu/~dbj/
Computer Science Department           Phone: [phone]
Carnegie Mellon University            Fax: [phone]
5000 Forbes Avenue
Pittsburgh, PA  15213-3891

-----------------------------------------------------------------------------
IETF Mobile IP Working Group Mailing List - Archives:  software.watson.ibm.com
Unsubscribe:	unsubscribe mobile-ip		(as message body, not subject)
Direct all administrative requests to [e-mail]