Question

NetIron CER 2048FX (IronWare : Version 6.0.0hT183 )

  • 31 July 2019
  • 7 replies
  • 138 views

In the instructions for setting up

netiron-6300a-switchingguide.pdf and netiron-05900a-switchingguide.pdf
It is written to make LAG do the following:

Configuring the primary port for a LAG
In previous versions of the Multi-Service IronWare software, the lowest number port was assigned as
the primary port in a LAG or LACP configuration. In version 03.7.00 and later, the primary port must be
explicitly assigned using the primary-port command.
To designate the primary port for the static LAG "blue", use the following command.

device(config)# lag blue static
device(config-lag-blue)# primary-port 3/2
Syntax: [no] primary-port slot/port

Once a primary port has been configured for a LAG, all configurations that apply to the primary port
are applied to the other ports in the LAG.

Configuring load sharing type
Individual LAGs can be configured to perform load sharing over the ports in the LAG using either a
hash based or per packet method, as shown in the following.

device(config)# lag blue static
device(config-lag-blue)# trunk-type hash-based
Syntax: [no] trunk-type hash-based | per-packet

my config:

lag "BRASDownlink" dynamic id 7
ports ethernet 1/41 to 1/42
primary-port 1/41
deploy
port-name "BRAStoP09" ethernet 1/41
port-name "BRAStoP10" ethernet 1/42
!

I do not have such teams. Who knows why?
Only these:

SSH@NetIron CER 2048FX(config-lag-BRASDownlink)
acl-mirror-port
clear
cls
deploy
disable
enable
end
exit
force-primary-port-mac-use
lacp-port-priority
lacp-timeout
monitor
no
physical-port
port-name
ports
primary-port
quit
sflow-forwarding
sflow-subsampling
show
trunk-threshold
update-lag-name
write

7 replies

Userlevel 1
If I remember correctly, this command is only available on the XMR/MLX/MLXE, the CER/CES is hash-based by default.
If I remember correctly, this command is only available on the XMR/MLX/MLXE, the CER/CES is hash-based by default.

SSH@NetIron CER 2048FX#sh lag BRASDownlink
Total number of LAGs : 4, 100/40g : 0
Total number of deployed LAGs : 4, 100/40g : 0
Total number of trunks created : 4 (60 available), 100/40g : 0 (0 available)
LACP System Priority / ID :1 / 748e.f862.ca01
LACP Long timeout :90, default: 90
LACP Short timeout :3, default: 3

=== LAG "BRASDownlink" ID 7 (dynamic Deployed) ===
LAG Configuration:
Ports: e 1/41 to 1/42
Port Count: 2
Primary Port: 1/41
Trunk Type: hash-based
forward_all_protocol: Disabled
LACP Key: 103

That is, I can not change to other options?

If I upload another firmware, for example XMR-MLX/06000g

Supported devices for R06.0.00g
The following devices are supported in this release:
ExtremeRouting XMR Series
ExtremeRouting MLX Series
ExtremeSwitching CES 2000 Series
ExtremeRouting CER 2000 Series
CER 2048FX
Userlevel 2
If I remember correctly, this command is only available on the XMR/MLX/MLXE, the CER/CES is hash-based by default.

That is correct. The CER is set to Hash-Based by default and this command is only available on the Chassis based systems.
Userlevel 1
I have 6.3a on MLX and CER, but these parameters are only visible on the MLX. Inside the documentation the CER limitation is buried somewhere.

code:
MLX:
mlxe(config-lag-vdx-core)#trunk-type ?
hash-based Hash based (default)
per-packet Per packet
(config)#load-balance ?
force-l4-hashing Force the use of L4 headers for trunk/ECMP hash
calculations (IPv4/v6)
hash-diversify Specify Trunk/ECMP Hash Diversification options
hash-rotate Control the rotation of the ECMP hash
mask Mask header fields for trunk/ECMP hash calculations
speculate-mpls-enet Use L2 header for trunk index calculation (LSR and
L2VPN PHP packets)
speculate-mpls-ip Use IP header for trunk index calculation (LSR and
L3VPN PHP packets)
symmetric Symmetric fields for trunk hash calculations
How to make it work on the CER 2048FX.
Need to buy a license or need to change сhassis ?
Userlevel 1
It is more a technical platform limit, not a license issue. Is there any reason, that you need per-packet load balancing?
I believe for the most use-cases, a hased-based load balancing is the better alternative.
LAG Balancing Problem:

PPPoE users <-> BRAS ( Routerboard ) <-> (4x1G LAG) <-> BGP Border (CER 2048FX.)

When the load rises to 1G, the packets are lost and the traffic does not go to other ports.

Reply