Several IRF (intelligent resilient
framework) supported switches of the same model can be interconnected to form a
fabric, in which each switch is a unit. The ports used to interconnect all the
units are called fabric ports, and the other ports that are used to connect the
fabric to users are called user ports. In this way, you can increase ports and
switching capability by adding devices to the fabric. In addition, reliability
of the system will be improved because the devices within the fabric can backup
each other. This feature brings you many advantages:
l
Realizes unified management of multiple devices.
Only one connection and one IP address are required to manage the entire
fabric. Therefore, management cost is reduced.
l
Enables you to purchase devices on demand and
expand network capacity smoothly. Protects your investment to the full extent
during network upgrade.
l
Ensures high reliability by N+1 redundancy,
avoids single point failure, and lessens service interruption.
Figure 1-1
Fabric
You can manage and maintain fabric topology
with the Fabric Topology Management (FTM) function. FTM on each unit exchanges
information with other units, including unit ID, fabric name, and the
authentication mode between units, by using a special kind of protocol packets.
It manages and maintains fabric topology according to the acquired information.
For example, when a new device is connected to a fabric, FTM will determine
whether it should establish a new fabric with the device according to the
obtained information.
The remote monitoring (RMON) configurations
of the devices in a fabric are the same. The RMON configuration performed on a
device of a fabric will be automatically synchronized to all devices in the fabric
if the configuration does not conflict with that of other devices in the
fabric.
If you configure the same entry in the same
RMON group for devices of a fabric to be different values, the value of the
RMON group entry will be the one configured by the device with the smallest
Unit ID when the devices are synchronized. Such a mechanism eliminates
configuration conflicts between the devices in a fabric.
After the configurations are consistent,
you can collect RMON history and statistics of any unit from any switch in the
fabric. For detailed information about RMON, refer to the SNMP-RMON Operation
part in H3C S5600 Series Ethernet Switches Operation Manual.
As the basis of the IRF function, the
fabric topology management (FTM) module manages and maintains the entire
topology of a fabric. The FTM module also implements the peer fabric port
detection function.
A device can join a fabric only when the
following conditions are met.
l
The number of the existing devices in the fabric
does not reach the maximum number of devices allowed by the fabric.
l
The fabric names of the device and the existing
devices in the fabric are the same.
l
The software version of the device is the same
as that of the existing devices in the fabric.
l
The device passes the security authentication if
security authentication is enabled in the fabric.
After a switch is powered on, the FTM
module releases device information of the switch through the fabric ports. The
device information includes Unit ID, CPU MAC, device type ID, fabric port
information, and all fabric configuration information. The device information
is released in the form of discovery packet (DISC). A new device can join a
fabric only when its DISC packets pass the authentication performed by the
existing devices in the fabric.
l
If a fabric port of a switch is connected to a
non-fabric port, the switch will not receive DISC packets from the peer. In
this case, the switch cannot join the fabric.
l
If the switch can receive DISC packets sent by
the peer, the FTM module determines whether peer sending ports correspond to
local receiving ports according to information in the packet. That is, if a
DISC packet received by the UP port of the switch is sent by the DOWN port of
the peer device, the packet is regarded legal. Otherwise, the packet is
regarded illegal and is discarded.
l
If the maximum number of devices allowed by the
fabric is reached, the devices in the fabric do not send DISC packets and
discard the received DISC packets. This prevents new devices from joining the
fabric.
l
After receiving a DISC packet from a directly
connected device, a device in a fabric checks whether the device information (that
is, the Fabric name and software version) contained in the packet and that of
its own are the same. If not, the received DISC packet is illegal and will be
discarded.
l
If authentication is enabled in the fabric, the
current device in the fabric authenticates received packets sent by new
directly connected devices. Packets that fail to pass the authentication will
be discarded.
The IRF Fabric peer detection function
outputs different prompt information according to the connection conditions of
the devices within the Fabric. You can refer to the following table to perform
corresponding adjustment and maintenance for the Fabric.
Table 1-1 Prompt
information analysis and solution
|
Prompt
|
Analysis
|
Solution
|
|
normal
|
it
indicates the fabric operates properly
|
The
“normal”, “temporary” and “redundance
port” information do not mean a device or a fabric operates improperly.
No measure is needed for any of these three types of information
|
|
temporary
|
it indicates the port status is changing
|
|
redundance port
|
it indicates the port is the redundant
port in fabric ring topology
|
|
connection
error
|
The port
matching three kinds of errors
|
Two fabric
ports of the same device (that is, the right port and the left port) are
connected.
|
Pull out
one end of the cable and connect it to a fabric port
of another switch.
|
|
The left
and right fabric ports of two devices are not connected in a crossed way.
|
Connect
the left and right ports of two devices in a crossed way.
|
|
A fabric
port of the local switch is connected to a non-fabric port, or is connected
to a fabric port that does not have fabric port function enabled.
|
Check the
types of the two interconnected ports on two sides. Make sure a fabric port
is only connected to ports of the same type and the fabric ports on both
sides are enabled with the fabric port function.
|
|
reached max units
|
it indicates that the maximum number of
units allowed by the current fabric is reached
|
Remove the new device or existing devices
in the fabric
|
|
different system name
|
it indicates the fabric name of the
device directly connected to the switch and the existing fabric name of the
fabric are not the same
|
Configure the fabric name of the new
device to be that of the fabric
|
|
different product version
|
indicates the software version of the
directly connected device and that of the current device are not the same
|
Make sure the software version of the new
device is the same as that of the fabric
|
|
auth failure
|
it indicates error occurs when the switch
authenticates a directly connected device. The error may occur if the IRF
fabric authentication modes configured for the both devices are not the same,
or the password configured does not match
|
Make sure the IRF fabric authentication
modes and the passwords configured for the both devices are the same
|
Up to eight devices
can be in an IRF fabric at a time.
FTM provides user interfaces. You can
configure unit IDs, fabric name by using a related command.
Table 1-2 Configure an IRF Fabric
1.3.2 Setting a Unit ID for a Switch
On the switches that support automatic
numbering, FTM will automatically number the switches to constitute an IRF
fabric, so that each switch has a unique unit ID in the fabric. You can use the
command in the following table to set unit IDs for switches. Make sure to set
different unit IDs for different switches in an IRF fabric. Otherwise, FTM will
automatically number the switches with the same unit ID.
Table 1-3 Set a unit ID for a switch
|
Operation
|
Command
|
Description
|
|
Enter system view
|
system-view
|
—
|
|
Set a unit ID for the switch
|
change self-unit to { unit-id | auto-numbering }
|
Optional
By default, the unit ID of a switch that
belongs to no IRF fabric is 1. The unit ID of a switch belonging to an IRF
fabric is assigned by FTM. Unit ID ranges from 1 to 8.
|
If you do not
enable the fabric port, you cannot change the unit ID of the local switch.
After an IRF fabric is established, you can
use the following command to change the unit IDs of the switches in the IRF
fabric.
Table 1-4 Set a unit ID to a new value
|
Operation
|
Command
|
Description
|
|
Enter system view
|
system-view
|
—
|
|
Set a unit ID to a new value
|
change unit-id unit-id1 to { unit-id2 | auto-numbering }
|
Optional
|
l
Unit IDs in an IRF fabric are not always
arranged in order of 1 to 8.
l
Unit IDs of an IRF fabric can be inconsecutive.
After you change the unit ID of switches,
the following operations are performed.
l
If the modified unit ID does not exist in the IRF
fabric, the system sets its priority to 5 and saves it in the unit Flash
memory.
l
If the modified unit ID is an existing one, the
system prompt you to confirm if you really want to change the unit ID. If you
choose to change, the existing unit ID is replaced and the priority is set to
5. Then you can use the fabric save-unit-id command to save the modified
unit ID into the unit Flash memory and clear the information about the existing
one.
l
If auto-numbering is selected, the system
sets the unit priority to 10. You can use the fabric save-unit-id
command to save the modified unit ID into the unit Flash memory and clear the
information about the existing one.
Priority is the
reference for FTM module to perform automatic numbering. The value of priority
can be 5 or 10. A smaller value represents a higher priority. Priority 5 means
the switch adopts manual numbering, and priority 10 means the switch adopts
automatic numbering.
After the configuration of numbering, you
can use the following command in the table to save the local unit ID in the
unit Flash memory. When you restart the switch, it can load the unit ID
configuration automatically.
Table 1-5 Save the unit ID of each unit
in the IRF fabric
|
Operation
|
Command
|
Description
|
|
Save the unit ID of each unit in the IRF
fabric
|
fabric save-unit-id
|
Optional
|
An S5600 series switch has two Cascade
ports, which are used to connect the devices within a Fabric. It is necessary
to enable the Fabric function for the Cascade ports and use local UP port to
connect peer DOWN port or use local DOWN port to connect peer UP port.
Otherwise, the Fabric cannot function normally.
Table 1-6 Specify
a fabric port
|
Operation
|
Command
|
Description
|
|
Enter system view
|
system-view
|
—
|
|
Specify the fabric port of a switch
|
fabric-port interface-type interface-number enable
|
Optional
|
l
Establishing an IRF system requires a high
consistency of the configuration of each device. Hence, before you enable the
fabric port, do not perform any configuration for the port, and do not enable
some functions that affect the IRF (such as TACACAS and BGP) for other ports or
globally. Otherwise, you cannot enable the fabric port. For detailed restrictions
refer to the error information output by devices.
l
As shutting down a fabric port directly may
cause the fabric being removed and error messages, do not perform such
operations.
l
To remove a fabric, you can simply remove the
cables used to form the fabric or disable fabric using the undo fabric-port
enable command.
l
You can shut down/bring up a port after you
disable the fabric feature on the port.
You can assign a unit name to a switch by
performing the operations listed in Table 1-7.
Table 1-7 Assign a unit name to a switch
|
Operation
|
Command
|
Description
|
|
Enter system view
|
system-view
|
—
|
|
Assign a unit name to a switch
|
set unit unit-id name unit-name
|
Required
|
Only the switches with the same IRF fabric
name can form an IRF fabric.
Table 1-8 Assign a fabric name
to a switch
|
Operation
|
Command
|
Description
|
|
Enter system view
|
system-view
|
—
|
|
Assign a fabric name to the switch
|
sysname sysname
|
Optional
By default, the IRF fabric name is H3C.
|
When an IRF fabric operates
normally, you can regard the whole fabric as a single device and perform
configuration on it. Multiple switches constitute an IRF fabric. Therefore,
data transmission and simultaneous program execution among the switches may
cause the IRF fabric in a busy situation. When you configure the IRF fabric,
you may receive a prompt “Fabric system is busy, please try
later…” which indicates the fabric system does not perform your
configuration properly. In this case, you need to verify your previous configuration
or perform your configuration again.
Following completion of the above
configuration, you can execute the display command in any view to view
device management and verify the settings. And you can execute the reset
command to clear the FTM statistics.
Table 1-9 Display and debug FTM
|
Operation
|
Command
|
Description
|
|
Display
the information about an IRF fabric
|
display
irf-fabric [ status ]
|
These commands
can be executed in any view.
|
|
Display
the topology information of an IRF fabric
|
display
ftm { information | topology-database
}
|
|
Display
RMON statistics of a specified unit in an IRF fabric
|
display
rmon statistics unit unit-id
|
|
Display
RMON history data of a specified unit in an IRF fabric
|
display
rmon history unit unit-id
|
|
Clear the
FTM statistics
|
reset ftm
statistics
|
Execute
this command in user view
|
Configure unit ID, unit name, IRF fabric
name for four switches to enable them to form an IRF fabric.
The configuration details are as follows:
l
Unit IDs: 1, 2, 3, 4
l
Unit names: unit 1, unit 2, unit 3, unit 4
l
Fabric name: hello
Figure 1-2
Network diagram for forming an IRF fabric
1)
Configure Switch A.
# Configure the unit ID as 1.
<H3C> system-view
[H3C] change unit-id 1 to 1
# Configure the unit name as unit 1.
[H3C] set unit 1 name unit1
# Configure the fabric name as hello.
[H3C] sysname hello
2)
Configure Switch B.
# Configure the unit ID as 2.
<H3C> system-view
[H3C] change unit-id 1 to 2
# Configure the unit name as unit 2.
[H3C] set unit 1 name unit2
# Configure the fabric name as hello.
[H3C] sysname hello
Configurations on Switch C and Switch D are
similar with the above configurations.
