Traffic passthrought to link aggregation

[Marco2]

Hello everyone,

I am trying to configure two HP5500 with dynamic bridge aggregation, everything seems ok but when tried to generate traffic the maximus allowed on link aggregation is 1 Gb, follows the switch configuration:

interface Bridge-Aggregation3
port link-type hybrid
undo port hybrid vlan 1
port hybrid vlan 2 tagged
port hybrid pvid vlan 2
link-aggregation mode dynamic

interface GigabitEthernet2/0/27
port link-mode bridge
port link-type hybrid
undo port hybrid vlan 1
port hybrid vlan 2 tagged
port hybrid pvid vlan 2
port link-aggregation group 3
#
interface GigabitEthernet2/0/28
port link-mode bridge
port link-type hybrid
undo port hybrid vlan 1
port hybrid vlan 2 tagged
port hybrid pvid vlan 2
port link-aggregation group 3
#
interface GigabitEthernet2/0/31
port link-mode bridge
port access vlan 2
#
interface GigabitEthernet2/0/32
port link-mode bridge
port access vlan 2

The configuration of the other switch is the same, switch are connected using eht 27-28, the eth 31-32 are connected to device used to generate traffic.

Checking the bridge aggregation the link is UP with speed at 2Gb

[HP]display interface Bridge-Aggregation3 brief
The brief information of interface(s) under bridge mode:
Link: ADM - administratively down; Stby - standby
Speed or Duplex: (a)/A - auto; H - half; F - full
Type: A - access; T - trunk; H - hybrid
Interface Link Speed Duplex Type PVID Description
BAGG3 UP 2G(a) F(a) H 2

[HP]display link-aggregation summary
Aggregation Interface Type:
BAGG -- Bridge-Aggregation, RAGG -- Route-Aggregation
Aggregation Mode: S -- Static, D -- Dynamic
Loadsharing Type: Shar -- Loadsharing, NonS -- Non-Loadsharing
Actor System ID: 0x8000, 7848-592d-0ac6

AGG AGG Partner ID Select Unselect Share
Interface Mode Ports Ports Type
-------------------------------------------------------------------------------
BAGG3 D 0x8000, b8af-67dd-4296 2 0 Shar

Aggregation Interface: Bridge-Aggregation3
Aggregation Mode: Dynamic
Loadsharing Type: Shar
System ID: 0x8000, 7848-592d-0ac6
Local:
Port Status Priority Oper-Key Flag
--------------------------------------------------------------------------------
GE2/0/27 S 32768 3 {ACDEF}
GE2/0/28 S 32768 3 {ACDEF}
Remote:
Actor Partner Priority Oper-Key SystemID Flag
--------------------------------------------------------------------------------
GE2/0/27 1 32768 3 0x8000, b8af-67dd-4296 {ACDEF}
GE2/0/28 2 32768 3 0x8000, b8af-67dd-4296 {ACDEF}

[HP]display interface Bridge-Aggregation3
Bridge-Aggregation3 current state: UP
IP Packet Frame Type: PKTFMT_ETHNT_2, Hardware Address: 7848-592d-0af7
Description: Bridge-Aggregation3 Interface
2Gbps-speed mode, full-duplex mode
Link speed type is autonegotiation, link duplex type is autonegotiation
PVID: 2
Port link-type: hybrid
Tagged VLAN ID : 2
Untagged VLAN ID : none
Last clearing of counters: Never
Last 300 seconds input: 60 packets/sec 4152 bytes/sec 0%
Last 300 seconds output: 60 packets/sec 4161 bytes/sec 0%
Input (total): 75748200 packets, 113872049811 bytes
75457006 unicasts, 288485 broadcasts, 2701 multicasts
Input (normal): 75748192 packets, - bytes
75457006 unicasts, 288485 broadcasts, 2701 multicasts
Input: 8 input errors, 0 runts, 0 giants, 0 throttles
8 CRC, 0 frame, - overruns, 0 aborts
- ignored, - parity errors
Output (total): 74454778 packets, 111914902652 bytes
74165344 unicasts, 286729 broadcasts, 2705 multicasts, 0 pauses
Output (normal): 74454778 packets, - bytes
74165344 unicasts, 286729 broadcasts, 2705 multicasts, 0 pauses
Output: 0 output errors, - underruns, - buffer failures
0 aborts, 0 deferred, 0 collisions, 0 late collisions
0 lost carrier, - no carrier

Any suggestions?

Thank you in advance
Marco

[Ivan_B · Accepted solution]

Hi @Marco2 !

The load-sharing mechanism in a BAGG is per-flow, not per packet. When a switch has a packet that should go out a BAGG, the switch uses certain (configurable) load-sharing algorithm that defines which physical link from all BAGG member ports will be used.

You can choose one of the following criteria or any combination for load sharing:
• Source/Destination MAC addresses
• Source/Destination service port numbers
• Ingress ports
• Source/Destination IP addresses

Alternatively, you can let the system automatically choose link-aggregation load-sharing criteria based
on packet types (Layer 2, IPv4, or IPv6 for example).

That's why the result that you will get is highly dependent on your testing methodology.  

 In Layer 2 aggregate interface view, the switch supports the following load sharing criteria and combinations:

• Load-sharing criteria automatically determined based on the packet type
• Source IP address
• Destination IP address
• Source MAC address
• Destination MAC address
• Destination IP address and source IP address
• Destination MAC address and source MAC address

like in this example:

system-view
interface Bri3
link-aggregation load-sharing mode destination-ip destination-mac source-up source-mac

Alternatively, you may use global load-sharing mode command, not the BAGG-specific. It will affect all BAGGs on the switch and it's got more options as sharing criteria:

• Load-sharing criteria automatically determined based on the packet type
• Source IP address
• Destination IP address
• Source MAC address
• Destination MAC address
• Source IP address and destination IP address
• Source IP address and source port
• Destination IP address and destination port
• Source IP address, source port, destination IP address, and destination port
• Any combination of incoming port, source MAC address, and destination MAC address

example:

system-view
link-aggregation load-sharing mode destination-ip destination-port

Taking all this into consideration, try to make a test that will generate many traffic flows with different attributes like source IP, destination IP, source MAC, destination MAC etc. Try different load-sharing modes, but keep in mind that load-sharing mode set under the BAGG takes precedence over the global one. 

And there is one specific of Comware IRF that is not very obvious - if you are running IRF stack, keep in mind that by default if a traffic enters an IRF Slot and needs to go out a BAGG that is configured on both Slots, it will ALWAYS use outgoing ports on this Slot. This is made in order to avoid IRF links overutilization. This behavior can be changed:

system-view
undo link-aggregation load-sharing mode local-first

Hope this helps!