Section 15.1.1.2 – Westermo RedFox Series User Manual
Page 325
Westermo OS Management Guide
Version 4.17.0-0
good scalability characteristics as the network is segmented into different broad-
cast domains.
Although technically feasible, it is strongly recommend to separate ring coupling
and routing, localizing the distinct functions in dedicated WeOS units, i.e., do not
use RiCo nodes also as routers.
Note
In some cases using RiCo nodes as routers makes sense. To ensure correct
operation of the RiCo node, the CPU bandwidth is reduced by default, i.e.,
when ”cpu-bandwidth-limit” is set to ”auto” (
) on a WeOS
unit configured for FRNT Ring Coupling or Multi-link Dual-Homing. This in
turn reduces routing performance.
This automatic reduction of CPU bandwidth can be overridden by changing
the CPU bandwidth limit setting (
15.1.1.2
Ring Coupling Hello Interval
RiCo nodes in the same FRNT ring exchange Hello messages as part of the active
uplink election process, see also
. These Hello messages are trans-
mitted every 100 ms by default. The hello interval can be fine tuned – a lower
value gives faster failover, but may have an adverse effect on the CPU usage.
When the CPU usage increases RiCo nodes may not be able to send Hello mes-
sages and will time out. This can lead to unpredictable performance and loss of
connectivity.
It is recommended that all RiCo nodes within an FRNT ring are configured with
the same Hello interval. If there are RiCo nodes with different Hello interval in an
FRNT ring, the protocol will default to the highest interval announced by any RiCo
node, i.e., a RiCo node’s effective hello interval may differ from its configured
hello interval. E.g., if you wish to transition from using ”hello-time 100” to
”hello-time 80”, all RiCo nodes will use interval 100 ms until all RiCo node’s in
the FRNT ring has been configured with interval 80 ms.
➞ 2015 Westermo Teleindustri AB
325