When configuring IRF fabric, go to these
sections for information you are interested in:
l
Introduction
to IRF
l
IRF
Fabric Configuration
l
Displaying
and Maintaining IRF Fabric
l
IRF
Fabric Configuration Example
1.1 Introduction
to IRF
Intelligent Resilient Framework (IRF), a
feature particular to H3C S5600 series switches, is a new technology for
building the core of a network. This feature allows you to build an IRF fabric
by interconnecting several S5600 series switches to provide more ports for
network devices and improve the reliability of your network.
I. Topology and connections of an IRF
fabric
An IRF fabric typically has a ring topology
structure. As shown in Figure
1-1, each S5600 switch uses two special ports on the rear panel
to connect with two other switches in the fabric. The two ports are called
fabric ports in general, UP port and DOWN port respectively; the other ports of
the switch, which are available for connections with users or devices outside
the fabric, are called user ports.
Figure 1-1
A schematic diagram of an IRF fabric
A correctly built IRF fabric features the
following:
l
Multiple S5600 series switches are interconnected
through their fabric ports.
l
Given a switch, its UP port is connected to the DOWN
port of another switch, and its DOWN port is connected to the UP port of a
third one.
Port connection mode for S5600 series ring
topology IRF fabric is shown in Figure 1-2.
Figure 1-2 Port connection mode for
S5600 series ring topology IRF fabric
IRF fabric also supports bus topology,
which has the same requirements as the ring topology. The difference is that each
of the switches across the bus connection is connected with the other switches
through only one fabric port, as shown in Figure 1-3.
Figure 1-3 Network diagram for IRF fabric with a bus topology
II. Fabric ports
On an S5600 series Ethernet switch, only the
two cascade ports on its rear panel can be configured as the fabric ports. The
two cascade ports are:
l
UP port: Cascade 1/2/1
l
DOWN port: Cascade 1/2/2
III. FTM
As the basis of the IRF function, the
Fabric Topology Management (FTM) program manages and maintains the entire
topology of a fabric.
With fabric ports configured, the FTM program
releases device information of the device 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 discovery packet (DISC).
After receiving the packet, the peer device
will analyze the packet. A device can form a fabric with the peer or 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 (up to eight
devices can form a fabric).
l
The fabric name 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.
In case IRF
automatic fabric is enabled, even if the software version of the local device
is inconsistent with that used on the device in the fabric, you can still add a
device to the fabric by automatic downloading and loading of the software.
IV. IRF fabric detection
Forming a fabric requires a high
consistency of connection modes between the devices and device information. Without
all the requirements for forming a fabric being met, a fabric cannot be formed.
The FTM program detects the necessary
conditions for forming a fabric one by one and displays the detection results.
You can use the display ftm information command to view the detection information
for the fabric, checking the running status of the fabric or analyzing the
problems. Table 1-1
lists the status and solution of the problems.
Table 1-1
Status and solution
|
Status
|
Analysis
|
Solution
|
|
normal
|
—
|
These three kinds of information do not
mean a device or a fabric operates improperly. No measure is needed for any
of them.
|
|
temporary
|
|
redundance port
|
|
connection error
|
Indicates three kinds of port matching
errors may occur.
|
Two fabric ports of the same device (that
is, the UP port and the DOWN port) are connected.
|
Pull out one end of the cable and connect
it to a fabric port of another switch.
|
|
The UP and DOWN fabric ports of the
devices are not connected in a crossed way.
|
Connect the UP and DOWN ports of two
devices in a crossed way.
|
|
A fabric port of the local switch is
connected to a fabric port that does not have fabric port function enabled.
|
Make sure that the fabric ports on both
sides are enabled with the fabric port function.
|
|
reached max units
|
The maximum number of units allowed by
the current fabric is reached. You will fail to add new devices to the fabric
in this case.
|
Remove the new device or existing devices
in the fabric.
|
|
different system name
|
The fabric name of the device directly
connected to the switch and the existing fabric name of the fabric are not
the same.
|
Modify the fabric name of the new device
to be that of the fabric.
|
|
different product version
|
The software version of the directly
connected device and that of the current device are not the same.
|
Update the software version to make sure
the software version of the new device is the same as that of the fabric.
|
V. IRF automatic fabric
If the software version and fabric name of
the local device are inconsistent with those of the device in the IRF fabric,
the local device cannot be added to the fabric. In this case, you have to manually
download and load the software, and then restart the device, or manually change
the fabric name to add the device to the fabric. H3C S5600 series switches
provide the IRF automatic fabric function, which enables the device to
automatically download the software and change the fabric name, thus reducing
the manual maintenance workload.
With IRF automatic fabric enabled, if inconsistency
in software version or fabric name occurs when a switch is added to a fabric,
the system automatically performs the following operations:
l
If the software version of the local device is
inconsistent with that of the device in the fabric, the system automatically
initiates a download request to the device with the smallest unit ID in the
fabric and downloads the software used by the device in the current fabric to
the local device. Then the device will automatically restart and be added to
the fabric.
l
If the fabric name of the local device is
inconsistent with that of the device in the fabric, the system automatically ignores
the inconsistency check of the fabric name and adds the device to the fabric.
Then the system automatically synchronizes the configurations to the device
with the smallest unit ID and changes the fabric name.
With the above operations completed, the device
can be added to the fabric and work normally.
Caution:
l
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.
l
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.
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.
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 integrating the routing tables
of all the devices. Then each slave synchronizes this forwarding table from the
master and takes it as the basis for layer 3 forwarding.
In this way, the forwarding table entries
of each device in the fabric can be consistent. Even if the master fails, other
devices can use the forwarding table synchronized from the master to perform
layer 3 forwarding, thus ensuring the accuracy of forwarding path. After re-electing
the master, the fabric will restart routing update.
III. DLA
As a new link aggregation mode, Distributed
Link Aggregation (DLA) can improve fault tolerance and redundancy backup of
user networks.
Link aggregation enables you to configure
ports on the same device as an aggregation port group, avoiding network
interruptions resulted from single port failure. Based on link aggregation, DLA
provides a more reliable solution, with which you can select ports on different
devices to form an aggregation port group. In this way, single port failure can
be avoided and network reliability can be greatly improved, because the fabric
can communicate with the destination network through ports on other devices in
case a single device fails.
1.2 IRF
Fabric Configuration
Complete the following tasks to configure IRF fabric:
1.2.2 Specifying
the Fabric Port of a Switch
Follow these steps to specify a fabric port:
|
To do…
|
Use the command…
|
Remarks
|
|
Enter system view
|
system-view
|
—
|
|
Specify the fabric port of a switch
|
fabric-port interface-type interface-number enable
|
Required
Not specified by default
|
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 configure
some functions that affect the IRF (such as TACACS and VLAN-VPN) for other
ports or globally. Otherwise, you cannot enable the fabric port. For detailed
restrictions refer to the error information output by devices.
l
After an IRF fabric is established successfully,
do not enable the burst function on any device within the IRF Fabric. For
introduction to the burst function, refer to the QoS&QoS Profile
part of the manual.
l
To split a fabric, you can simply remove the
cables used to form the fabric or disable the fabric using the undo
fabric-port enable command.
l
If you need to configure an IRF fabric as a DHCP
server, configure the UDP Helper function in the fabric at the same time to
ensure that the client can successfully obtain an IP address. (Since this
configuration can be automatically synchronized to the entire fabric, you can
perform it on only one unit.) For the configuration of the UDP Helper function,
refer to the UDP Helper part of this manual.
1.2.3 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 by default, 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.
Follow these steps to set a unit ID for a switch:
|
To do…
|
Use the command…
|
Remarks
|
|
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.
Follow these steps to set a unit ID to a
new value:
|
To do…
|
Use the command…
|
Remarks
|
|
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 program to perform automatic numbering. The value of priority
can be 5 or 10. Priority 5 means the switch adopts manual numbering, and
priority 10 means the switch adopts automatic numbering. Manual numbering has
a higher priority than 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.
Follow these steps to save the unit ID of
each unit in the IRF fabric:
|
To do…
|
Use the command…
|
Remarks
|
|
Save the
unit ID of each unit in the IRF fabric
|
fabric
save-unit-id
|
Optional
|
1.2.4 Assigning
a Unit Name to a Switch
You can assign
a unit name to a switch by performing the operations listed in the following
table.
Follow these
steps to assign a unit name to a switch:
|
To do…
|
Use the command…
|
Remarks
|
|
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.
Follow these steps to assign a fabric name to a switch:
|
To do…
|
Use the command…
|
Remarks
|
|
Enter system view
|
system-view
|
—
|
|
Assign a fabric name to the switch
|
sysname sysname
|
Optional
By default, the IRF fabric name is H3C.
|
I. Configuration prerequisites
l
Make sure that the Flash of the newly added
device has enough space to download software used on the device in the fabric.
l
Configure the fabric port for the newly added
device.
II. Configuration procedure
Follow these
steps to configure IRF automatic fabric for a switch:
|
To do…
|
Use the command…
|
Remarks
|
|
Enter
system view
|
system-view
|
—
|
|
Configure
IRF automatic fabric for a switch
|
fabric
member-auto-update software enable
|
Required
Disabled
by default
|
Caution:
l
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.
l
After you configure the IRF automatic fabric
function on Slave, execute the save command to save the configurations
as soon as possible. Otherwise, the device may synchronize the configurations
from Master and restart repeatedly because the configurations on the device are
lost after the device automatically downloads the software and restarts.
When fabric works
normally, you can configure the whole fabric as an individual device. As a
fabric is comprised of multiple devices, busy working state may occur due to
data transmission between devices or synchronous execution of programs. When
you perform an operation, if you receive a prompt “Fabric system is busy,
please try later…”, which indicates that the fabric does not
execute your configuration properly, you need to verify your configuration or
reconfigure the previous operation.
|
To do…
|
Use the command…
|
Remarks
|
|
Display the information about an IRF
fabric
|
display irf-fabric [ status ]
|
Available in any view
|
|
Display the topology information of an
IRF fabric
|
display ftm { information | topology-database }
|
|
Clear the FTM statistics
|
reset ftm statistics
|
Available in user view
|
Configure unit ID, unit name, and IRF fabric
name for four switches to enable them to form an IRF fabric as shown in Figure 1-4.
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 1-4 Network diagram for forming an
IRF fabric
1)
Configure Switch A.
# Configure fabric ports.
<H3C> system-view
[H3C] fabric-port Cascade 1/2/1
enable
[H3C] fabric-port Cascade 1/2/2
enable
# 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 fabric ports.
<H3C> system-view
[H3C] fabric-port Cascade 1/2/1
enable
[H3C] fabric-port Cascade 1/2/2
enable
# Set the unit ID to 2.
[H3C] change unit-id 2 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.
