Chronological Thread 

< Chronological >      < Thread >

---

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Friday 27 June 2003 08:47, Bala Kanagasabai wrote:
> Hi all,
>
> I am trying out a test case where a server is multihomed with two
> transport addresses in an endpoint. The additional IP addresses are
> created using the following:
>
> ifconfig eth0:0 192.168.0.200
> route add -host 192.168.0.200 dev eth0:0
>
> ifconfig eth0:1 192.168.0.201
> route add -host 192.168.0.201 dev eth0:1
>
>
> The server endpoint is denoted as:
>
> Echo Server= [192.168.0.200, 192.168.0.201:7]

Bala,

one thing to note here is: you will not be able to make real use
of multi-homing, unless you do have two physically distinct paths,
and you can make one of them fail (e.g. by unplugging), while the
other one is still active (e.g. you can ping the second address fine,
while the physical path to the first address is disconnected).

In your scenario, say you have two hosts A and B with two addresses each,
say A1 and A2, and B1 and B2. Normally, A1 will talk to B1, and A2 to B2.
If the physical layer and routing are correct, you can ping A1-B1 while
path 2 is disconnected, and A2-B2 while path 1 is disconnected, where as
the other adress pairs will not work. You might want to check that first.

Another solution would be modification of the default route, after the
SCTP path has failed. Note that by default, if both destination addresses
are in the same subnet (maybe your own subnet) your system will likely always
use the same interface (and thus probably the same source address) for
outgoing
packets. This *may not get* you real dual-homing ! Ethereal is your friend :->

Anyway, if the primary path fails (physically), SCTP will *retransmit* to the
secondary (which should work), and will increase failure counters for the
primary path. After a while (say, 1 minute or so), the primary path will
become unreachable (which should give you a notification) and you should make
another, active path (probably path 2 if you only have 2) new primary.
Then you have an SCTP failover.

You will need to read the socket-api draft in order to find out,
how to code this with the socket api. See
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpsocket-06.txt
for reference. And yes, you need to read all of it... :-)

Regards,
Andreas

- --
Dipl.-Ing. Andreas Jungmaier
Computer Networking Technology Group
University of Duisburg-Essen
http://www.exp-math.uni-essen.de/~ajung
My PGP Key-ID: D3824AC0

My PGP Key-Fingerprint: 228C 7C2C 3381 2DD4 9998 2812 5C0B 0B04 D382 4AC0
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQE+/EIRXAsLBNOCSsARAsaPAJ43a1/pi/EwW8LMd8MwH1y10s3q5ACgs5S8
SLRpAauKqXK0hMpeffKH46c=
=U+pM
-----END PGP SIGNATURE-----

---

• [SCTP-Discussion] Multihoming connections using socketapi, Bala Kanagasabai, 06/27/2003
  • Re: [SCTP-Discussion] Multihoming connections using socketapi, Andreas Jungmaier, 06/27/2003

---

Archive powered by MHonArc 2.6.19+.