HPE FlexFabric 12900E User manual

HPE FlexFabric
12900E & 12900 & 7900 Switch Series
Virtual Technologies Configuration Guide
HPE FlexFabric 12900E Switch Series
HPE FlexFabric 12900 Switch Series
HPE FlexFabric
7900 Switch Series
P
art number: 5200-4954a
Software
version: Release 2710 and later
Document version: 6W100-2018426

© Copyright 2018 Hewlett Packard Enterprise Development LP
The information contained herein is subject to change without notice. The only warranties for Hewlett Packard
Enterprise products and services are set forth in the express warranty statements accompanying such
products and services. Nothing herein should be construed as constituting an additional warranty. Hewlett
Packard Enterprise shall not be liable for technical or editorial errors or omissions contained herein.
Confidential computer software. Valid license from Hewlett Packard Enterprise required for possession, use, or
copying. Consistent with FAR 12.211 and 12.212, Commercial Computer Software, Computer Software
Documentation, and Technical Data for Commercial Items are licensed to the U.S. Government under vendor’s
standard commercial license.
Links to third-party websites take you outside the Hewlett Packard Enterprise website. Hewlett Packard
Enterprise has no control over and is not responsible for information outside the Hewlett Packard Enterprise
website.
Acknowledgments
Intel®, Itanium®, Pentium®, Intel Inside®, and the Intel Inside logo are trademarks of Intel Corporation in the
United States and other countries.
Microsoft® and Windows® are trademarks of the Microsoft group of companies.
Adobe® and Acrobat® are trademarks of Adobe Systems Incorporated.
Java and Oracle are registered trademarks of Oracle and/or its affiliates.
UNIX® is a registered trademark of The Open Group.

i
Contents
Configuring an IRF fabric ···································································1
About IRF···································································································································1
IRF network model ················································································································1
IRF benefits ·························································································································2
Basic concepts ·····················································································································3
IRF network topology ·············································································································5
Master election ·····················································································································5
Interface naming conventions··································································································6
File system naming conventions·······························································································6
Configuration synchronization··································································································7
Multi-active handling procedure································································································7
MAD mechanisms ·················································································································9
Restrictions and guidelines: IRF configuration ················································································· 14
Hardware compatibility with IRF····························································································· 14
Software requirements for IRF ······························································································· 15
IRF fabric size ···················································································································· 15
Member ID restrictions ········································································································· 15
Candidate IRF physical interfaces ·························································································· 15
Transceiver modules and cables selection for IRF····································································· 16
IRF port connection ············································································································· 16
IRF physical interface configuration restrictions and guidelines ····················································17
Feature compatibility and configuration restrictions with IRF························································18
IRF tasks at a glance·················································································································· 18
Planning the IRF fabric setup ······································································································· 19
Setting up an IRF fabric··············································································································· 20
Assigning a member ID to each IRF member device·································································· 20
Specifying a priority for each member device············································································20
Binding physical interfaces to IRF ports··················································································· 20
Saving configuration to the next-startup configuration file···························································· 21
Connecting IRF physical interfaces························································································· 21
Setting the operating mode to IRF mode·················································································· 21
Accessing the IRF fabric······································································································· 22
Configuring MAD ······················································································································· 22
Restrictions and guidelines for MAD configuration ····································································· 22
Configuring LACP MAD········································································································ 22
Configuring BFD MAD·········································································································· 23
Configuring ARP MAD·········································································································· 27
Configuring ND MAD ··········································································································· 30
Excluding service interfaces from the shutdown action upon detection of multi-active collision··········· 31
Recovering an IRF fabric ······································································································ 32
Optimizing IRF settings for an IRF fabric ························································································ 32
Changing the member ID of a member device ··········································································32
Changing the priority of a member device ················································································ 33
Adding physical interfaces to an IRF port················································································· 33
Bulk-configuring basic IRF settings for a member device ···························································· 34
Enabling IRF auto-merge······································································································35
Configuring a member device description················································································· 35
Configuring the IRF bridge MAC address················································································· 35
Enabling software auto-update for software image synchronization···············································37
Setting the IRF link down report delay····················································································· 38
Removing an interface card that has IRF physical interfaces ·······················································38
Replacing an interface card that has IRF physical interfaces ······················································· 38
Display and maintenance commands for IRF··················································································· 39
IRF configuration examples·········································································································· 39
Example: Configuring an LACP MAD-enabled two-chassis IRF fabric ··········································· 39
Example: Configuring a BFD MAD-enabled two-chassis IRF fabric··············································· 41
Example: Configuring an ARP MAD-enabled two-chassis IRF fabric ·············································44

ii
Example: Configuring an ND MAD-enabled two-chassis IRF fabric··············································· 46
Example: Restoring standalone mode····················································································· 49
Configuring MDCs··········································································52
About MDC······························································································································· 52
MDC advantages ················································································································ 52
MDC applications················································································································ 52
Default MDC and non-default MDCs ······················································································· 53
Interface management ········································································································· 53
Restrictions and guidelines: MDC configuration ··············································································· 53
MDC configuration tasks at a glance······························································································ 53
Creating an MDC······················································································································· 54
Starting an MDC························································································································ 54
Assigning hardware resources to an MDC ······················································································ 55
About assigning hardware resources ······················································································ 55
Assigning physical interfaces and LPUs to MDCs······································································ 55
Specifying a CPU weight for an MDC······················································································ 57
Specifying a memory space percentage for an MDC·································································· 58
Accessing an MDC····················································································································· 58
Returning to the default MDC ······································································································· 59
Deleting an MDC ······················································································································· 59
Display and maintenance commands for MDC················································································· 60
MDC configuration examples········································································································ 60
Example: Configuring MDCs (in standalone mode)···································································· 60
Example: Configuring MDCs (in IRF mode)·············································································· 64
Document conventions and icons ······················································72
Conventions······························································································································ 72
Network topology icons ··············································································································· 73
Support and other resources·····························································74
Accessing Hewlett Packard Enterprise Support················································································ 74
Accessing updates····················································································································· 74
Websites ··························································································································· 74
Customer self repair ············································································································ 75
Remote support ·················································································································· 75
Documentation feedback ······································································································ 75
Index···························································································76

1
Configuring an IRF fabric
About IRF
The Intelligent Resilient Framework (IRF) technology virtualizes multiple physical devices at the
same layer into one virtual fabric to provide data center class availability and scalability. IRF
virtualization technology offers processing power, interaction, unified management, and
uninterrupted maintenance of multiple devices.
IRF network model
Figure 1 shows an IRF fabric that has two devices, which appear as a single node to the upper-layer
and lower-layer devices.
Figure 1 IRF application scenario
As shown in Figure 2, Device A and Device B form a two-chassis IRF fabric. The fabric has four
MPUs (one active and three standbys), and two times the number of interface cards that a single
device provides. The IRF fabric manages the physical and software resources of Device A and
Device B in a centralized manner.
IP network
IRF fabric
IP network
IRF link Simplified to
Master Subordinate

2
Figure 2 Two-chassis IRF fabric implementation schematic diagram
IRF benefits
IRF provides the following benefits:
•
Simplified topology and easy management—An IRF fabric appears as one node and is
accessible at a single IP address on the network. You can use this IP address to log in at any
member device to manage all the members of the IRF fabric. In addition, you do not need to run
the spanning tree feature among the IRF members.
•
1:N redundancy—In an IRF fabric, one member acts as the master to manage and control the
entire IRF fabric. All the other members process services while backing up the master. When
the master fails, all the other member devices elect a new master from among them to take over
without interrupting services.
•
IRF link aggregation—You can assign several physical links between neighboring members to
their IRF ports to create a load-balanced aggregate IRF connection with redundancy.
•
Multichassis link aggregation—You can use the Ethernet link aggregation feature to
aggregate the physical links between the IRF fabric and its upstream or downstream devices
across the IRF members.
•
Network scalability and resiliency—Processing capacity of an IRF fabric equals the total
processing capacities of all the members. You can increase ports, network bandwidth, and
processing capacity of an IRF fabric simply by adding member devices without changing the
network topology.
IRF link
An IRF
fabric is
formed.
Master
(Member ID=1)
Subordinate
(Member ID=2)
Active MPU
Device A
(Member ID=1) Device B
(Member ID=2)
IRF physical
interfaces IRF physical
interfaces
IRF-port 2 IRF-port 1
Network
interfaces
Network
interfaces
IRF
Standby MPU
Active MPU
Standby MPU
IRF’s master MPU
IRF’s standby MPU
IRF’s standby MPU
IRF’s standby MPU

3
Basic concepts
Operating mode
The device operates in one of the following modes:
•
Standalone mode—The device cannot form an IRF fabric with other devices.
•
IRF mode—The device can form an IRF fabric with other devices.
IRF member roles
IRF uses two member roles: master and standby(called subordinate throughout the documentation).
When devices form an IRF fabric, they elect a master to manage and control the IRF fabric, and all
the other devices back up the master. When the master device fails, the other devices automatically
elect a new master. For more information about master election, see "Master election."
IRF member ID
An IRF fabric uses member IDs to uniquely identify and manage its members. This member ID
information is included as the first part of interface numbers and file paths to uniquely identify
interfaces and files in an IRF fabric. Two devices cannot form an IRF fabric if they use the same
member ID. A device cannot join an IRF fabric if its member ID has been used in the fabric.
MPU roles
Each IRF member device has one or two MPUs. The following are MPU roles:
Role
Description
Master MPU Active MPU of the master device. It is also called the global active MPU. You
configure and manage the entire IRF fabric at the CLI of the global active
MPU.
Active MPU
Active MPU on each member device. An active MPU performs the following
tasks:
•Manages the local device, including synchronizing configuration with the
local standby MPU, processing protocol packets, and creating and
maintaining route entries.
•Processes IRF-related events, such as master election and topology
collection.
Standby MPU
For the master MPU, all other MPUs are standby MPUs, including active
MPUs on subordinate devices.
If a member device has two MPUs, the MPU backing up the local active MPU
is the local standby MPU from the perspective of the member device.
Member priority
Member priority determines the possibility of a member device to be elected the master. A member
with higher priority is more likely to be elected the master.
IRF port
An IRF port is a logical interface that connects IRF member devices. Every IRF-capable device has
two IRF ports.
In standalone mode, the IRF ports are named IRF-port 1 and IRF-port 2.
In IRF mode, the IRF ports are named IRF-port n/1 and IRF-port n/2, where nis the member ID of the
device. The two IRF ports are referred to as IRF-port 1 and IRF-port 2 in this book.
To use an IRF port, you must bind a minimum of one physical interface to it. The physical interfaces
assigned to an IRF port automatically form an aggregate IRF link. An IRF port goes down when all its
IRF physical interfaces are down.

4
IRF physical interface
IRF physical interfaces connect IRF member devices and must be bound to an IRF port. They
forward traffic between member devices, including IRF protocol packets and data packets that must
travel across IRF member devices.
IRF split
IRF split occurs when an IRF fabric breaks up into multiple IRF fabrics because of IRF link failures,
as shown in Figure 3. The split IRF fabrics operate with the same IP address. IRF split causes
routing and forwarding problems on the network. To quickly detect a multi-active collision, configure
a minimum of one MAD mechanism (see "Configuring MAD").
Figure 3 IRF split
IRF merge
IRF merge occurs when two split IRF fabrics reunite or when two independent IRF fabrics are united,
as shown in Figure 4.
Figure 4 IRF merge
MAD
An IRF link failure causes an IRF fabric to split in two IRF fabrics operating with the same Layer 3
settings, including the same IP address. To avoid IP address collision and network problems, IRF
uses multi-active detection (MAD) mechanisms to detect the presence of multiple identical IRF
fabrics, handle collisions, and recover from faults.
IRF domain ID
One IRF fabric forms one IRF domain. IRF uses IRF domain IDs to uniquely identify IRF fabrics and
prevent IRF fabrics from interfering with one another.
As shown in Figure 5, IRF fabric 1 contains Device Aand Device B, and IRF fabric 2 contains Device
C and Device D. Both fabrics use the LACP aggregate links between them for MAD. When a
member device receives an extended LACPDU for MAD, it checks the domain ID to determine
whether the packet is from the local IRF fabric. Then, the member device can handle the packet
correctly.
=
IRF link
Device A Device B
IRF fabric
Device A Device B
IRF fabric 1 IRF fabric 2
+
IRF link
Device A Device B
Device A Device B
IRF fabric 1 IRF fabric 2 IRF fabric
+=

5
Figure 5 A network that contains two IRF domains
IRF network topology
The 12900E, 12900 or 7900 switch series supports only two-chassis IRF fabrics in daisy-chain
topology, as shown in Figure 6.
Figure 6 Daisy-chain topology
No intermediate devices are allowed between neighboring members.
Master election
Master election occurs each time the IRF fabric topology changes in the following situations:
•
The IRF fabric is established.
•
The master device fails or is removed.
•
The IRF fabric splits.
•
Independent IRF fabrics merge.
Device A Device B
IRF fabric 1
(domain 10)
IRF link
Core network
IRF fabric 2
(domain 20)
IRF link
Device C Device D
Access network
IRF-Port1IRF-Port2
IRF

6
NOTE:
Master election does not occur
when split IRF fabrics merge. For information about the master
device of the merged IRF fabric, see "Failure recovery."
Master election selects a master in descending order:
1. Current master, even if a new member has higher priority.
When an IRF fabric is being formed, all members consider themselves as the master. This rule
is skipped.
2. Member with higher priority.
3. Member with the longest system uptime.
Two members are considered to start up at the same time if the difference between their startup
times is equal to or less than 10 minutes. For these members, the next tiebreaker applies.
4. Member with the lowest CPU MAC address.
For the setup of a new IRF fabric, the subordinate devices must reboot to complete the setup after
the master election.
For an IRF merge, devices must reboot if they are in the IRF fabric that fails the master election.
Interface naming conventions
In standalone mode:
A physical interface is numbered in the slot-number/subslot-number/interface-index format.
For example, set the link type of FortyGigE 1/0/1 to trunk, as follows:
<Sysname> system-view
[Sysname] interface fortygige 1/0/1
[Sysname-FortyGigE1/0/1] port link-type trunk
In IRF mode:
A physical interface is numbered in the chassis-number/slot-number/subslot-number/interface-index
format. The chassis-number argument represents the IRF member ID.
For example, FortyGigE 1/2/0/1 represents the first port in slot 2 on member device 1. Set its link
type to trunk, as follows:
<Sysname> system-view
[Sysname] interface fortygige 1/2/0/1
[Sysname-FortyGigE1/2/0/1] port link-type trunk
File system naming conventions
In standalone mode, you can use the storage device name to access the file system of the active
MPU. To access the file system of the standby MPU, use the name in the
slotslot-number#storage-device-name format.
On a multichassis IRF fabric, you can use the storage device name to access the file system of the
global active MPU. To access the file system of a global standby MPU, use the name in the
chassismember-ID#slotslot-number#storage-device-name format.
For more information about storage device naming conventions, see Fundamentals Configuration
Guide.
For example:
•
To create and access the test folder under the root directory of the flash memory on the global
active MPU:

7
<Master> mkdir test
Creating directory flash:/test... Done.
<Master> cd test
<Master> dir
Directory of flash:/test
The directory is empty.
1048576 KB total (651568 KB free)
•
To create and access the test folder under the root directory of the flash memory in slot 0 on
member device 1:
<Master> mkdir chassis1#slot0#flash:/test
Creating directory chassis1#slot0#flash:/test... Done.
<Master> cd chassis1#slot0#flash:/test
<Master> dir
Directory of chassis1#slot0#flash:/test
The directory is empty.
1048576 KB total (651568 KB free)
Configuration synchronization
IRF uses a strict running-configuration synchronization mechanism. In an IRF fabric, all MPUs obtain
and run the running configuration of the global active MPU. Configuration changes are automatically
propagated from the global active MPU to the remaining MPUs. The configuration files of these
MPUs are retained, but the files do not take effect. The subordinate devices use their own startup
configuration files on their respective local active MPU only after these devices are removed from the
IRF fabric.
As a best practice, back up the next-startup configuration file on a device before adding the device to
an IRF fabric as a subordinate.
A subordinate device's next-startup configuration file might be overwritten if the master and the
subordinate use the same file name for their next-startup configuration files. You can use the backup
file to restore the original configuration after removing the subordinate from the IRF fabric.
For more information about configuration management, see Fundamentals Configuration Guide.
Multi-active handling procedure
The multi-active handling procedure includes detection, collision handling, and failure recovery.
Detection
IRF provides MAD mechanisms by extending LACP, BFD, ARP, and IPv6 ND.
MAD identifies each IRF fabric with a domain ID and an active ID (the member ID of the master). If
multiple active IDs are detected in a domain, MAD determines that an IRF collision or split has
occurred.
For more information about the MAD mechanisms and their application scenarios, see "MAD
mechanisms."
Collision handling
When detecting a multi-active collision, MAD disables all IRF fabrics except one from forwarding
data traffic by placing them in Recovery state. The IRF fabrics placed in Recovery state are called
inactive IRF fabrics. The IRF fabric that continues to forward traffic is called the active IRF fabric.

8
LACP MAD and BFD MAD use the following process to handle a multi-active collision:
1. Compare the number of members in each fabric.
2. Set all fabrics to the Recovery state except the one that has the most members.
3. Compare the member IDs of the masters if all IRF fabrics have the same number of members.
4. Set all fabrics to the Recovery state except the one that has the lowest numbered master.
5. Shut down all service interfaces in the Recovery-state fabrics except for the following service
interfaces:
Service interfaces automatically excluded from being shut down by the system.
Service interfaces specified by using the mad exclude interface command.
ARP MAD and ND MAD use the following process to handle a multi-active collision:
1. Compare the member IDs of the masters in the IRF fabrics.
2. Set all fabrics to the Recovery state except the one that has the lowest numbered master.
3. Take the same action on the service interfaces in Recovery-state fabrics as LACP MAD and
BFD MAD.
Failure recovery
To merge two split IRF fabrics, first repair the failed IRF link and remove the IRF link failure.
After the failed IRF link between two split IRF fabrics is recovered, log in to the inactive IRF fabric to
reboot its member devices if the system requires you to do so. If the IRF auto-merge feature is
enabled, the inactive IRF member devices will automatically reboot and a manual reboot is typically
not required. After these member devices join the active IRF fabric as subordinate devices, the IRF
merge is complete, as shown in Figure 7. The service interfaces that have been shut down by MAD
automatically restore their original state.
If you mistakenly reboot the active IRF fabric, the member devices in the IRF fabric join the inactive
IRF fabric as subordinate devices. You must use the mad restore command to restore the original
state of the service interfaces that have been shut down by MAD.
Figure 7 Recovering the IRF fabric
If the active IRF fabric fails before the IRF link is recovered (see Figure 8), use the mad restore
command on the inactive IRF fabric to recover the inactive IRF fabric. This command brings up all
service interfaces that were shut down by MAD. After the IRF link is repaired, merge the two parts
into a unified IRF fabric.
IP network
IP network
IRF fabric 1
(Active)
IRF fabric 2
(Recovery)
IP network
IP network
IRF fabric 1
(Active) IRF fabric 2
(Recovery) IRF
fabric
IP network
IP network
After the IRF link
is recovered IRF merge

9
Figure 8 Active IRF fabric fails before the IRF link is recovered
MAD mechanisms
IRF provides MAD mechanisms by extending LACP, BFD, ARP, and IPv6 ND.
Table 1 compares the MAD mechanisms and their application scenarios.
Table 1 Comparison of MAD mechanisms
MAD
mechanism
Advantages Disadvantages Application
scenarios
LACP MAD
•Detection speed is fast.
•Runs on existing
aggregate links without
requiring MAD-dedicated
physical links or Layer 3
interfaces.
Requires an intermediate
device that supports
extended LACP.
Link aggregation is used
between the IRF fabric
and its upstream or
downstream device.
BFD MAD
•Detection speed is fast.
•Intermediate device, if
used, can come from any
vendor.
Requires MAD dedicated
physical links and Layer 3
interfaces, which cannot be
used for transmitting user
traffic.
•No special
requirements for
network scenarios.
•If no intermediate
device is used, this
mechanism is only
suitable for IRF
fabrics that have
only two members
that are
geographically
IRF
fabric 1
(Active)
IRF fabric 2
(Recovery)
IP network
IP network
IRF fabric 2
(Recovery)
IP network
IP network
IRF fabric 1
fails before
the IRF link
is recovered.
IRF fabric 1
fails
because of
physical
problems
IRF fabric 2
(Active)
IP network
IP network
IRF fabric 1
fails
because of
physical
problems
IRF
fabric
IP network
IP network
Repair the
IRF link and
IRF fabric 1,
and finish
IRF merge
Execute the mad
restore command
on IRF fabric 2

10
MAD
mechanism
Advantages Disadvantages Application
scenarios
close to one
another.
ARP MAD
•No intermediate device is
required.
•Intermediate device, if
used, can come from any
vendor.
•Does not require MAD
dedicated ports.
•Detection speed is
slower than BFD MAD
and LACP MAD.
•The spanning tree
feature must be
enabled if common
Ethernet ports are used
for ARP MAD links.
Spanning tree-enabled
non-link aggregation
IPv4 network scenarios if
common Ethernet ports
are used.
ND MAD
•No intermediate device is
required.
•Intermediate device, if
used, can come from any
vendor.
•Does not require MAD
dedicated ports.
•Detection speed is
slower than BFD MAD
and LACP MAD.
•The spanning tree
feature must be
enabled.
Spanning tree-enabled
non-link aggregation
IPv6 network scenarios.
LACP MAD
As shown in Figure 9, LACP MAD has the following requirements:
•
Every IRF member must have a link with an intermediate device.
•
All the links form a dynamic link aggregation group.
•
The intermediate device must be a device that supports extended LACP.
The IRF member devices send extended LACPDUs that convey a domain ID and an active ID. The
intermediate device transparently forwards the extended LACPDUs received from one member
device to all the other member devices.
•
If the domain IDs and active IDs sent by all the member devices are the same, the IRF fabric is
integrated.
•
If the extended LACPDUs convey the same domain ID but different active IDs, a split has
occurred. LACP MAD handles this situation as described in "Collision handling."

11
Figure 9 LACP MAD scenario
BFD MAD
BFD MAD detects multi-active collisions by using BFD.
You can use common or management Ethernet ports for BFD MAD.
If management Ethernet ports are used, BFD MAD has the following requirements:
•
An intermediate device is required and each IRF member device must have a BFD MAD link to
the intermediate device.
•
Each member device is assigned a MAD IP address on the master's management Ethernet
port.
If common Ethernet ports are used, BFD MAD has the following requirements:
•
If an intermediate device is used, each member device must have a BFD MAD link to the
intermediate device. If no intermediate device is used, all member devices must have a BFD
MAD link to each other. As a best practice, use an intermediate device to connect IRF member
devices if the IRF fabric has more than two member devices. Afull mesh of IRF members might
cause broadcast loops.
•
Ports on BFD MAD links are assigned to a VLAN or Layer 3 aggregate interface used for BFD
MAD. Each member device is assigned a MAD IP address on the VLAN interface or Layer 3
aggregate interface.
The BFD MAD links, BFD MAD VLAN, and BFD MAD VLAN interface (or Layer 3 aggregate
interface) must be dedicated. Do not use the BFD MAD links, BFD MAD VLAN, or BFD MAD VLAN
interface (or Layer 3 aggregate interface) for any other purposes.
Intermediate device
Master Subordinate
IRF
fabric
Internet
Customer
premise
network
IRF link
Common traffic path
LACP MAD traffic path
LACP-enabled dynamic
link aggregation
LACP-enabled dynamic
link aggregation

12
NOTE:
•
The MAD addresses identify the member devices and must belong to the same subnet.
•
Of all management Ethernet ports on an IRF fabric, only the master's management Ethernet port
is accessible.
Figure 10 shows a typical BFD MAD scenario that uses an intermediate device. On the intermediate
device, assign the ports on the BFD MAD links to the same VLAN.
Figure 11 shows a typical BFD MAD scenario that does not use an intermediate device.
With BFD MAD, the master attempts to establish BFD sessions with other member devices by using
its MAD IP address as the source IP address.
•
If the IRF fabric is integrated, only the MAD IP address of the master takes effect. The master
cannot establish a BFD session with any other member. If you execute the display bfd
session command, the state of the BFD sessions is Down.
•
When the IRF fabric splits, the IP addresses of the masters in the split IRF fabrics take effect.
The masters can establish a BFD session. If you execute the display bfd session
command, the state of the BFD session between the two devices is Up.
Figure 10 BFD MAD scenario with an intermediate device
Figure 11 BFD MAD scenario without an intermediate device
ARP MAD
ARP MAD detects multi-active collisions by using extended ARP packets that convey the IRF
domain ID and the active ID.
You can use common or management Ethernet ports for ARP MAD.
If management Ethernet ports are used, ARP MAD must work with an intermediate device. Make
sure the following requirements are met:
•
Connect a management Ethernet port on each member device to the intermediate device.
•
On the intermediate device, you must assign the ports used for ARP MAD to the same VLAN.
Device
Master Subordinate
IRF fabric
IRF link
192.168.1.2/24 192.168.1.3/24
BFD MAD link
BFD MAD link
Master Subordinate
IRF fabric
IRF link
BFD MAD link
VLAN 2
192.168.1.2/24 VLAN 2
192.168.1.3/24

13
If common Ethernet ports are used, ARP MAD can work with or without an intermediate device.
Make sure the following requirements are met:
•
If an intermediate device is used, connect each IRF member device to the intermediate device,
as shown in Figure 12. In this situation, data links can be used. On the intermediate device,
assign the ports on the ARP MAD links to the same VLAN. Run the spanning tree feature
between the IRF fabric and the intermediate device.
•
If no intermediate device is used, connect each IRF member device to all other member
devices. In this situation, IRF links cannot be used for ARP MAD.
Each IRF member compares the domain ID and the active ID in incoming extended ARP packets
with its domain ID and active ID.
•
If the domain IDs are different, the extended ARP packet is from a different IRF fabric. The
device does not continue to process the packet with the MAD mechanism.
•
If the domain IDs are the same, the device compares the active IDs.
If the active IDs are different, the IRF fabric has split.
If the active IDs are the same, the IRF fabric is integrated.
Figure 12 ARP MAD scenario
ND MAD
ND MAD detects multi-active collisions by using NS packets to transmit the IRF domain ID and the
active ID.
ND MAD can work with or without an intermediate device. Make sure the following requirements are
met:
•
If an intermediate device is used, connect each IRF member device to the intermediate device,
as shown in Figure 13. In this situation, data links can be used. On the intermediate device,
Device
Master Subordinate
IRF
fabric
Internet
Customer
premise
network
IRF link
Common traffic path
Extended ARP traffic path
STP domain (all devices
must run the spanning
tree feature)

14
assign the ports on the ND MAD links to the same VLAN. Run the spanning tree feature
between the IRF fabric and the intermediate device.
•
If no intermediate device is used, connect each IRF member device to all other member
devices. In this situation, IRF links cannot be used for ND MAD.
Each IRF member device compares the domain ID and the active ID in incoming NS packets with its
domain ID and active ID.
•
If the domain IDs are different, the NS packet is from a different IRF fabric. The device does not
continue to process the packet with the MAD mechanism.
•
If the domain IDs are the same, the device compares the active IDs.
If the active IDs are different, the IRF fabric has split.
If the active IDs are the same, the IRF fabric is integrated.
Figure 13 ND MAD scenario
Restrictions and guidelines: IRF configuration
Hardware compatibility with IRF
An HPE FlexFabric 12900E switch can form an IRF fabric only with devices of the same model. In
addition, an HPE FlexFabric 12900E switch that uses type-H service modules cannot form an IRF
fabric with an HPE FlexFabric 12900E switch that uses type-F service modules.
An HPE FlexFabric 12900 switch can form an IRF fabric only with devices of the same model.
An HPE FlexFabric 7900 switch can form an IRF fabric only with devices of the same model.
Device
Master Subordinate
IRF
fabric
Internet
Customer
premise
network
IRF link
Common traffic path
Extended ND traffic path
STP domain (all devices
must run the spanning
tree feature)

15
If the global active MPU switches to an MPU that has a higher number of management Ethernet
ports than the current global active MPU, the number of accessible management Ethernet ports
does not change. Only the first n management Ethernet ports are accessible on the new global
active MPU. The integer n equals the number of management Ethernet ports on the original global
active MPU.
In other situations, all management Ethernet ports on the global active MPU are accessible.
Software requirements for IRF
All IRF member devices must run the same software image version. Make sure the software
auto-update feature is enabled on all member devices.
IRF fabric size
An HPE FlexFabric 12900E IRF fabric can contain a maximum of two member devices.
An HPE FlexFabric 12900 IRF fabric can contain a maximum of two member devices.
An HPE FlexFabric 7900 IRF fabric can contain a maximum of two member devices.
Member ID restrictions
The following matrix shows the IRF member ID and hardware compatibility:
Hardware
IRF member ID compatibility
HPE FlexFabric 12901E Switch Chassis (JH951A)
with type-H service modules 1 or 2
HPE FlexFabric 12902E Switch Chassis (JH345A)
with type-H service modules 1 or 2
HPE FlexFabric 12904E Switch Chassis (JH262A)
and HPE FlexFabric 12904E TAA-compliant
Switch Chassis (JH266A) with type-H service
modules
1 or 2
HPE FlexFabric 12908E Switch Chassis (JH255A)
and HPE FlexFabric 12908E TAA-compliant
Switch Chassis (JH259A) with type-H service
modules
1 or 3
HPE FlexFabric 12916E Switch Chassis (JH103A)
and HPE FlexFabric 12916E TAA-compliant
Switch Chassis (JH109A) with type-H service
modules
1 or 5
HPE FlexFabric 12900E with type-F service
modules 1 or 2
HPE FlexFabric 12900 with type-F service
modules 1 or 2
HPE FlexFabric 7900 with type-F service modules 1 or 2
Candidate IRF physical interfaces
Use the following ports on the 12900E, 12900 or 7900 switch series for IRF links:
•
10GBase-T Ethernet ports operating at 10 Gbps.

16
•
SFP+ ports operating at 10 Gbps.
•
QSFP+ ports operating at 40 Gbps.
•
QSFP28 ports operating at 100 Gbps.
•
CFP2 ports operating at 100 Gbps.
•
CXP ports operating at 100 Gbps.
Ports enabled with the MACsec feature cannot be used as IRF physical interfaces. For more
information about the MACsec feature, see Security Configuration Guide.
The split (the using tengige command) and combination (the using fortygige command) of
a 40-GE QSFP+ port require a card reboot. As a best practice to avoid topology change, complete
these operations before you add the device to an IRF fabric. When you perform these operations in
an IRF fabric, make sure you understand the impact on the IRF fabric topology.
Transceiver modules and cables selection for IRF
When you select transceiver modules and cables, follow these restrictions and guidelines:
•
Use Category 6A(or above) twisted-pair cables to connect 10GBase-T Ethernet ports in a short
distance.
•
Use SFP+, QSFP+, QSFP28, CFP2, or CXP transceiver modules and fibers to connect SFP+,
QSFP+, QSFP28, CFP2, or CXP ports for a long-distance connection.
•
Use SFP+, QSFP+, QSFP+ to SFP+, or QSFP28 DAC cables to connect SFP+, QSFP+, or
QSFP28 ports for a short-distance connection.
•
Use CXP optical cables to connect CXP ports for a short-distance connection.
•
The transceiver modules at the two ends of an IRF link must be the same type.
For more information about the transceiver modules and DAC cables, see the switch installation
guide and HPE Comware-Based Devices Transceiver Modules User Guide.
NOTE:
The
transceiver modules and DAC cables available for the switch are subject to change over time.
For the most up
-to-date list of transceiver modules and DAC cables, contact your Hewlett Packard
Enterprise sales representative.
IRF port connection
When you connect two neighboring IRF members, follow these restrictions and guidelines:
•
You must connect the physical interfaces of IRF-port 1 on one member to the physical
interfaces of IRF-port 2 on the other.
•
For high availability, bind multiple physical interfaces on different service modules to an IRF
port. You can bind a maximum of 16 physical interfaces to an IRF port.
Figure 14 Connecting IRF physical interfaces
IRF-port 1IRF-port 2
IRF fabric