the default config of my RB configures different types of interfaces, and I would like to understand what they mean. Could not find much information in the manual.
The interface list gives me
# NAME TYPE MTU L2MTU MAX-L2MTU
0 R ether1-gateway ether 1500 1598 4074
1 S ether2 ether 1500 1598 4074
2 S ether3 ether 1500 1598 4074
3 S ether4 ether 1500 1598 4074
4 RS ether5 ether 1500 1598 4074
5 RS ether6-master-local ether 1500 1598 2028
6 S ether7-slave-local ether 1500 1598 2028
7 S ether8-slave-local ether 1500 1598 2028
8 S ether9-slave-local ether 1500 1598 2028
9 S ether10-slave-local ether 1500 1598 2028
10 R bridge-local bridge 1500 1598
ether1-gateway is my WAN gateway. bridge-local is a virtual interface, that combines ether2..ether6.
The bridge creates a broadcast-domain, so all broadcasts are forwarded to all links of the bridge.
Traffic with individual destination is only forwarded to the link where that destination is expected (correct?). But what does the slave “S” mean, and what is specific to ether6-master-local and ether7..10-slave-local ?
And how/when should I use the different types in my network ?
thanks for your reply and the reference.
Let me clarify the questions:
Interfaces ether1..5 have their “Switch”-attribute set to switch1, ether6..10 have it set to switch2. This corresponds to the two switch chips of the router. Is the following correct?:
the association between interfaces and switches[1|2] if fixed and cannot be modified
without further configuration, no switching takes place, all ports are equal and routed through the RB CPU
switching between two or more ports can be enabled by setting the same master port. In my configuration, ether7..10 have the same master port, so packets between these ports are switched, bypassing the CPU, rather than routed (via the RB CPU).
the master port and the switched ports must connect to the same switch chip.
So after reading “Interface-Case Studies-Switch Chip Features”, this is my current understanding of port switching, I hope it is correct (please comment!). But what does it mean that “Interfaces for which the ‘master’ port is specified become inactive”? In my config, ether6 is the master of ether7..10 and the hosts connected to it can communicate just like any other.
Now what about bridging? Accroding to the manual, “the bridge feature allows the interconnection of hosts connected to separate LANs (…) as if they were attached to a single LAN”. Refering to my configuration, is the following correct?:
ether6..10 are configured as a port switched group, because ether6 is master of the others. So these ports appear as one single LAN.
all other interfaces represent individual LANs. The “bridge-local” specifies a bridge that brings these different LANs together, plus ether6 as the master the switching group.
Summarising this, and ignoring additional features (like rule definition) for the moment, port switching is a good (fast) method when ports connected to the same switch chip shall form a LAN. The bridge is the solution if ports that are not connected to the same switch chip shall appear as a singel LAN.
Essentially, yes. The diagram above shows some of the internal bottlenecks in the hardware links.
Exactamundo.
Interfaces with Master-port set cannot have IP addresses applied to them and cannot be configured manually. You, in essence, lose control of them but the master port itself can be configured.