SCTP failover

From: Anatoly Khusid <anatoly.khusid@ulticom.com>
Date: Fri Feb 11 2005 - 12:29:36 EST

('binary' encoding is not supported, stored as-is) ('binary' encoding is not supported, stored as-is) Hello,

 I am using Linux SCTP implementation (LKSCTP) SLES9 (2.6.5-7.111.19-smp)
distribution.
 I have a client application that is sending data to a server on a remote
 machine. The machines are connected over two private LANs.
 When I disconnect a primary interface, I expect SCTP to start using an
 alternative LAN as soon as possible. Well, it takes about one minute for
 LKSCTP to detect that LAN is down, before it starts to transmit data
 messages on another LAN. I don’t have any data messages lost, but I have a
 1-minute delay during which time no data is received by a server, after
 about one minute, the data transmission resumes through the alternative
 interface.
 I am curious why it takes so long to detect a LAN failure? I am using all
 the defaults for SCTP provisioning. (In fact I used getsockopt() to verify
 that the defaults match SCTP specs).
 Based on this section in SCTP RFC I would expect the switchover to be in a
 matter of seconds the most. I am using Ulticom’s SCTP implementation and
 the switchover only takes about 1 second. Could anyone please shed some
 light on this?

 Section 6.4 of SCTP RFC 2960:
 Furthermore, when its peer is multi-homed, an endpoint SHOULD try to
 retransmit a chunk to an active destination transport address that is
 different from the last destination address to which the DATA chunk was
 sent.

 Thanks,

 Anatoly Khusid
 Ulticom Inc.
  Senior Software Engineer
Received on Fri Feb 11 12:54:15 2005