How Irf Works - H3C S5600 Series Operation Manual

Operation Manual – IRF Fabric
H3C S5600 Series Ethernet Switches
Caution:
You need to enable the IRF automatic fabric function on all the devices including the
newly added device in the fabric to enable the newly added device to download
software and discovery neighbors and thus be added to the fabric normally.
You are recommended to set the Unit ID of the switch with the software to be
downloaded to 1, and thus ensure that the candidate switch can download a correct
software version.

1.1.2 How IRF Works

When a fabric is established, the devices determine their respective roles in the fabric
by comparing their CPU MAC addresses. The device with the lowest CPU MAC
address is elected as the master and the other devices are slaves.
After the election, the fabric can operate normally. The following three functions of IRF
can provide simple configuration mode, enhanced network performance and perfect
redundancy backup mechanism for users.
I. DDM
Distributed Device Management (DDM) is a new device management mode provided
by IRF. In normal cases, a fabric can be considered as a single device. You can
manage the entire fabric by logging onto any device in the fabric with different logging
modes. The devices in the fabric synchronize their configurations by exchanging
packets, thus ensuring stability of the fabric.
FTM program uses Unit ID, or device ID to distinguish between the devices in a fabric
when you manage them. On initialization of the IRF function, each device considers its
Unit ID as 1 and after a fabric connection is established, the FTM program
automatically re-numbers the devices or you can manually configure the Unit ID of
them.
The master in a fabric collects the newest configurations of the user and the slaves
periodically synchronize the configurations from the master. In this way, the entire fabric
can operate with the same configurations.
II. DRR
Distributed Redundancy Routing (DRR) is used to implement redundancy routing
backup. The devices in a fabric run their independent routing protocols and maintain
their own routing tables. Unlike a common layer 3 switch, a fabric member does not
generate a layer 3 forwarding table to forward packets; instead, it uploads the routing
table to the master, which generates a forwarding table used by the entire fabric by
Chapter 1 IRF Fabric Configuration
1-5